Myobatrachus, Schlegel, 1850

Frost, Darrel R., Grant, Taran, Faivovich, Julián, Bain, Raoul H., Haas, Alexander, De Sá, Célio F. B. Haddad Rafael O., Channing, Alan, Wilkinson, Mark, Donnellan, Stephen C., Raxworthy, Christopher J., Campbell, Jonathan A., Blotto, Boris L., Moler, Paul, Drewes, Robert C., Nussbaum, Ronald A., Lynch, John D., Green, David M. & Wheeler, And Ward C., 2006, The Amphibian Tree Of Life, Bulletin of the American Museum of Natural History 2006 (297), pp. 1-291 : 1-291

Myobatrachus and Arenophryne to obtain

Myobatrachinae Schlegel In Gray, 1850b: 10. sister taxa because they both are head­first Type genus: Myobatrachus Schlegel, 1850 . burrowers in sandy soil (Tyler, 1989), with

196 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

all of the concomitant morphological features per Amazon Basin to the Amazonian slopes that are associated with this behavior. of the Andes.

CONTENT: Hemiphractus Wagler, 1828 . [348] NOBLEOBATRACHIA NEW TAXON CHARACTERIZATION AND DIAGNOSIS: None ETYMOLOGY: Noble (Gladwyn K. Noble) of the morphological characters in our anal­ 1 batrachos (Greek: frog), to note one of the ysis optimize on this branch because, as most influential herpetologists of the twenti­ these species are direct­developers and thereeth century and the father of modern inte­ fore were not studied by Haas (2003). Nevgrative herpetology. Although Noble died ertheless, Hemiphractus / Hemiphractidae is relatively young, at age 47 (Adler, 1989), his easily diagnosed by its wild appearance and contributions to amphibian systematics, life triangular skull. Like most basal species of history, comparative anatomy, and experi­ Meridianura, Hemiphractus exhibits direct mental biology remain important milestones. development and bears the developing em­ IMMEDIATELY MORE INCLUSIVE TAXON: bryos on the back until hatching, but unlike [318] Notogaeanura new taxon. species of Amphignathodontidae and Cryp­ SISTER TAXON: [319] Australobatrachia tobatrachidae, it does not have a dorsal new taxon. pouch in which to carry developing embryos RANGE: Coextensive with Anura , exclud­ (Noble, 1931). ing New Zealand and the Seychelles. SYSTEMATIC COMMENT : Hemiphractus has CONCEPT AND CONTENT : Nobleobatrachia is two pairs of bell­shaped gills in embryos, de­ a monophyletic group containing Hemi­ rived from branchial arches I and II (del Pino phractidae Peters, 1862, and [349] Meridi­ and Escobar, 1981; Mendelson et al., 2000), anura new taxon. as do members of Cryptobatrachidae and CHARACTERIZATION AND DIAGNOSIS: Opti­ Amphignathodontidae (except for Flectonomization of claw­shaped terminal phalanges tus pygmaeus ). This suggests that the char­ and intercalary elements is ambiguous, acter of bell­shaped gills optimizes on Merplaced on this branch only under accelerated idianura, with a reversal in Athesphatanura. optimization. Under delayed optimization, however, the characters appear convergently [349] MERIDIANURA NEW TAXON in Hemiphractidae and in Cladophrynia. The ETYMOLOGY: Meridianus (Greek: southern) character of bell­shaped gills optimizes on 1 anoura (Greek: tailless, i.e., frog), refer­ Meridianura, with a reversal at Athesphatan­ encing the South American center of distriura. Nevertheless, the bulk of evidence for bution of this worldwide group. the existence of this clade is molecular; see IMMEDIATELY MORE INCLUSIVE TAXON: appendix 5 for molecular synapomorphies. [348] Nobleobatrachia new taxon. The length of the 28S fragment likely be­ SISTER TAXON: Hemiphractidae Peters , comes much longer (greater than 740 bp) at 1862. this branch than found below this point (see appendix 3), although this must be confirmed RANGE: Coextensive with Anura , exclud­

ing New Zealand and the Seychelles. by examining the 28S fragment in Hemiphractus . CONCEPT AND CONTENT : Meridianura is a

monophyletic group containing [350] Bra­ FAMILY: HEMIPHRACTIDAE PETERS, 1862 chycephalidae Günther, 1858, and [366] Cla­

dophrynia new taxon. Hemiphractidae Peters, 1862: 146 . Type genus: CHARACTERIZATION AND DIAGNOSIS: No Hemiphractus Wagler, 1828 . morphological characters in our analysis op­ IMMEDIATELY MORE INCLUSIVE TAXON: timize on this branch, and no authors have [348] Nobleobatrachia new taxon. suggested morphological characters that SISTER TAXON: [349] Meridianura new tax­ would optimize here. Nevertheless, it is wellon. corroborated by molecular characters (see

RANGE: Panama; Pacific slopes of Colom­ appendix 5 for molecular synapomorphies). bia and northwestern Ecuador; Brazil, Co­ Meridianura is characterized by a length of lombia, Ecuador, Peru, and Bolivia in the up­ the 28S DNA fragment in excess of 740 bp

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 197

(appendix 3). This may be plesiomorphic, exhibits the further derived character of ovoshared with Sooglossidae and reversed in viviparity (Drewery and Jones, 1976). In ad­ Australobatrachia, but long 28S molecules dition, embryonic egg teeth have been reare characteristic nonetheless. ported for Brachycephalus and Eleutherodactylus (Duellman and Trueb, 1986; Pom­ [350] FAMILY: BRACHYCEPHALIDAE bal, 1999). Additional survey may find that GÜNTHER, 1858

this characteristic is synapomorphic for some Brachycephalina Günther, 1858a: 321. Type ge­ larger group; is likely coextensive with direct nus: Brachycephalus Fitzinger, 1826 . development in this group; and therefore is Cornuferinae Noble, 1931: 521. Type genus: possibly synapomorphic for the entire Bra­ Cornufer Tschudi, 1838 . chycephalidae. For molecular transformations Eleutherodactylinae Lutz, 1954: 157. Type genus: associated with this taxon see appendix 5.

Eleutherodactylus Duméril and Bibron, 1841 . S: We find nominal YSTEMATIC COMMENTS

IMMEDIATELY MORE INCLUSIVE TAXON: Eleutherodactylus (sensu lato—subtended by [349] Meridianura new taxon. branch 350) to be in the same situation as SISTER TAXON: [366] Cladophrynia new nominal ‘‘ Hyla ’’ prior to its partition by Faitaxon. vovich et al. (2005) into several tribes and RANGE: Tropical North and South Ameri­ many new genera—that of a gigantic and illca; Antilles. defined group where the enormity of the tax­ CONTENT: Adelophryne Hoogmoed and on and lack of understanding of its species Lescure, 1984; Atopophrynus Lynch and diversity has largely restricted taxonomic Ruiz­Carranza, 1982; Barycholos Heyer , work for the past 45 years to two individuals 1969; Brachycephalus Fitzinger, 1826 ; Dis­ (John D. Lynch and Jay M. Savage) 28. Nevchidodactylus Lynch , 1979; Craugastor ertheless, the current taxonomy of Eleuther­ Cope, 1862 (see Systematic Comments and odactylus (sensu lato, subdivided into the appendix 7); ‘‘ Eleutherodactylus ’’ Duméril taxa Craugastor , Euhyas , Eleutherodactylus and Bibron, 1841 (see Systematic Comments [sensu stricto], Pelorius , and Syrrhophus ) ex­ and appendix 7); ‘‘ Euhyas ’’ Fitzinger, 1843 tends from the work of Hedges (1989) in (see Systematic Comments and appendix 7); which he named Pelorius for the Eleuthero­ Euparkerella Griffiths, 1959 ; Geobatrachus dactylus inoptatus group and placed Tomo­ Ruthven, 1915; Holoaden Miranda­Ribeiro , dactylus as a synonym of Syrrhophus and his 1920; Ischnocnema Reinhardt and Lütken , enlarged Syrrhophus as a subgenus of 1862 ‘‘1861’’; ‘‘ Pelorius ’’ Hedges, 1989 (see Eleutherodactylus .

Systematic Comments and appendix 7); Hedges’ (1989) systematic arrangement Phrynopus Peters, 1873 ; Phyllonastes Heyer , was based on an allozymic study of six loci 1977; Phyzelaphryne Heyer, 1977 ; Syrrho­ (223 alleles) focused on West Indian species, phus Cope, 1878 (including Tomodactylus with a narrative discussion of evidence sup­ Günther, 1900; see Systematic Comments porting recognition of non­West Indian taxa. and appendix 7). In his UPGMA tree, Hedges’ Group I (native CHARACTERIZATION AND DIAGNOSIS: Bra­ Jamaican species, except E. nubicola ) apchycephalids are predominantly leaf­litter pears monophyletic. His Group II ( E. ricordii frogs with axillary amplexus and direct de­ group) obtained as polyphyletic, with two velopment (J.D. Lynch , 1971, 1973). None groups placed far from each other, one group of the morphological characters in our anal­ (paraphyletic to group I) and another group ysis optimized on this branch due to incom­ much more basal. Group III ( E. auriculatus plete taxon sampling in our morphological group) obtained as polyphyletic, with both a data set (from Haas, 2003, who restricted his basal and a ‘‘central’’ monophyletic group. study to groups with larvae). Nevertheless, Group IV ( E. inoptatus group) obtained as a Brachycephalidae is characterized by pos­

sessing very large terrestrial eggs and exhib­

28 G.A. Boulenger (1882), in his extraordinarily influto

.

iting direct development in all species so far

ential ‘‘Catalogue of the Batrachia Salientia’’, had examined (J.D. Lynch , 1971), with the ex­ deal with only about 50 species of what is now Eleuthception of Eleutherodactylus jasperi , which erodactylus (sensu lato). Life was much simpler then 198 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

monophyletic group. In the same paper a son to believe this will not worsen as sam­ Distance Wagner tree also obtained Group I pling density increases. In addition to the as monophyletic, Group II as polyphyletic, paraphyly of ‘‘ Eleutherodactylus ’’ (sensu Group III as polyphyletic, and IV as mono­ lato) with respect to Brachycephalus , disphyletic. cussed in ‘‘Results’’, we found Ischnocnema ,

After performing these analyses, Hedges Barycholos , and Phrynopus to be imbedded rejected the idea that any significant evolu­ within it, as was anticipated by Ardila­Rotionary meaning attached to these results, bayo (1979). and suggested that Groups I–IV are each As regards Ischnocnema, J.D. Lynch monophyletic on the basis of possessing (1972b: 9) noted its extreme resemblance to unique alleles (no overall analysis of pres­ species of the E. binotatus group and could ence–absence provided): Group I (Icdq2), not eliminate the possibility that Ischnocne­ Group II (PgmjB), Group III (Icdf1), Group IV ma represents ‘‘a single morphological di­ (Icdp5, Lgla1, Pgm 0). (This survey of loci was vergence of the binotatus group of Eleuthbased solely on Antillean taxa, without any erodactylus ’’. Our placement of E. binotatus sampling of the nominal subgenera Craugas­ and Ischnocnema quixensis as sister taxa tor or Syrrhophus , and with very limited supports that hypothesis (see below). sampling of mainland species of subgenus The sole basis for recognizing Phrynopus Eleutherodactylus .) Hedges then considered as distinct from ‘‘ Eleutherodactylus ’’ (sensu Group I and Group II together to form his lato or sensu stricto) is the absence of exsubgenus Euhyas ; the rationale for this uni­ panded digital discs (J.D. Lynch , 1975). Exfication was not provided. His Group III he panded discs are also absent in several speregarded as the E. auriculatus section of a cies of ‘‘ Eleutherodactylus ’’ (sensu stricto), presumed paraphyletic subgenus Eleuthero­ which J.D. Lynch (1994: 201) considered to dactylus (referred to later as Eleutherodac­ be evidence that ‘‘ Phrynopus are simply tylus [sensu stricto]), and Group IV he con­ Eleutherodactylus having greatly reduced sidered to be his monophyletic subgenus Pe­ digital tips’’. Our taxon sampling was inadlorius. In subsequent discussion, he noted equate to address the relationships among all that J.D. Lynch (1986) had provided a mor­ brachycephalids (i.e., eleutherodactylines) phological synapomorphy for a group that with unexpanded discs and provided only a Hedges had not examined, Craugastor (the minimal test of Phrynopus monophyly, but mandibular ramus of the trigeminal nerve ly­ our results leave little doubt that Phrynopus ing medial [deep] to the m. adductor man­ is nested within ‘‘ Eleutherodactylus ’’ (sensu dibulae externus superficialis, the ‘‘E’’ condition of Starrett in J.D. Lynch , 1986), which lato). Hedges also accepted as a subgenus. Hedges J.D. Lynch (1980) considered Barycholos briefly discussed why he rejected Savage’s to be most closely related to Eleutherodac­ (1987) contention that Tomodactylus and tylus nigrovittatus (then placed in the E. dis­ Syrrhophus are distantly related, and then re­ coidalis group but subsequently transferred garded them as synonymous (as Syrrhophus ) to the new E. nigrovittatus group by J.D. and considered Syrrhophus to be a subgenus Lynch , 1989). We did not sample any species of Eleutherodactylus . As with other authors of the E. nigrovittatus group in this study and before and since, Hedges provided no evi­ therefore did not test the assertion of a Bardence for the monophyly of Eleutherodac­ ycholos– E. nigrovittatus relationship directtylus (sensu lato) with respect to other eleuth­ ly. However, our finding (following Caraerodactyline genera. J.D. Lynch and Duell­ maschi and Pombal, 2001) that Barycholos man (1997) disputed some assignments to ternetzi is nested within ‘‘ Eleutherodactylus ’’ Euhyas , but otherwise accepted Hedges’ ar­ (sensu lato) is consistent with J.D. Lynch’s rangement. hypothesis. We also did not test the mono­

Our results showed Eleutherodactylus phyly of Barycholos , which is characterized

(sensu lato) to be rampantly nonmonophy­ by sternal architecture (primarily the occurletic (indicated below by quotation marks rence of a calcified sternal style; J.D. Lynch , surrounding the name), and there is no rea­ 1980), but the 3,200 km separation between

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 199

the only known species is strongly sugges­ subgenera is meager or lacking, several less tive that it may not be monophyletic. inclusive species groups are delimited by Adelophryne , Brachycephalus , and Eupar­ synapomorphy, and we anticipate that several kerella share the characteristic of reduction of these will be recognized formally as of phalanges in the fourth finger, a presumed knowledge increases. synapomorphy, but this does not prevent this As a preliminary step in this direction, we group from being imbedded within ‘‘ Eleuth­ take the action of treating all of the nominal erodactylus ’’ (sensu lato or sensu stricto) 29, subgenera of ‘‘ Eleutherodactylus ’’ (sensu nor are we aware of any other characters that lato) as genera. (As noted in the ‘‘Review of would exclude any of the other nominal gen­ Current Taxonomy’’, Crawford and Smith, era of Brachycephalidae (including former 2005, on the basis of molecular data, recently Eleutherodactylinae) from ‘‘ Eleutherodacty­ considered Craugaster a genus.) As dislus ’’ (sensu lato or sensu stricto). cussed later, this is only partially successful Given the extent of the demonstrated non­ inasmuch as it leaves ‘‘ Eleutherodactylus ’’, monophyly and lack of any evidence to dis­ ‘‘ Euhyas ’’, and ‘‘ Pelorius ’’ of dubious tinguish even a phenetic ‘‘ Eleutherodacty­ monophyly or even demonstrated polyphyly lus ’’ assemblage from other brachycephalid (denoted by the quotation marks). Neverthegenera, the only immediately available rem­ less, this illuminates the attendant problems edy, and the most scientifically conservative of brachycephalid relationships and leaves us action in that it enforces monophyly as the in an operationally healthier place than organizing principle of taxonomy, would be where we had been. That is, the extent of our to place all species of the former Eleuthero­ knowledge of monophyly is represented by dactylinae in a single genus (coextensive the recognition of Brachycephalidae and the with our Brachycephalidae ), for which the demonstrably monophyletic units within it, oldest available name is Brachycephalus Fit­ and other genera are merely provisional units zinger, 1826. But, as much as this appeals to of convenience. (See appendix 7 for new us in principle, we believe that, in this par­ combinations produced by these generic ticular case—where knowledge is so limited changes.) and species diversity is so great, and where Among the previous subgenera of we have sampled so few of the nominal gen­ ‘‘ Eleutherodactylus ’’ (sensu lato) we found era of Brachycephalidae (i.e., we have not [358] Syrrhophus to be monophyletic (tested sampled Adelophryne , Atopophrynus , Dis­ by inclusion of S. marnocki of the S. marchidodactylus , Euparkerella , Geobatrachus , nocki group of J.D. Lynch and Duellman, Holoaden , Phyllonastes , or Phyzelaphry­ 1997, and S. nitidus of the S. nitidus group ne)—the scientific payoff from enforcing of J.D. Lynch and Duellman, 1997; see apmonophyly is not worth the practical cost of pendix 5 for molecular synapomorphies). obscuring so much diversity under a single Our sole representative of the Antillean Eugeneric name and thereby concealing a con­ hyas (represented by E. planirostris of the E. siderable number of phylogenetic hypotheses ricordii group of J.D. Lynch and Duellman, that we would rather advertise in order to 1997) did not allow us to test the monophyly attract more work. Moreover, we strongly be­ of this taxon. lieve that progress in the scientific under­ In an admittedly weak test of monophyly, standing of these frogs will be achieved by we included two species of Eleutherodactypartitioning ‘‘ Eleutherodactylus ’’ into multi­ lus (sensu stricto), both of the E. binotatus ple monophyletic genera. Indeed, although group: E. binotatus and E. juipoca . Neverevidence for the monophyly of the nominal theless, we refuted the monophyly of

‘‘ Eleutherodactylus ’’ (sensu stricto) (and the 29 Complicating this, Adelophryne and Phyzelaphryne E. binotatus group), showing E. binotatus (Hoogmoed and Lescure, 1984) and at least some mem­ and E. juipoca to be more closely related to bers of the Eleutherodactylus diastema group (T. Grant, Ischnocnema and Brachycephalus , respec­

personal obs.) possess conspicuously pointed tips on the

tively. Although this finding was unanticigenera within former ‘‘ Eleutherodactylus ’’ (sensu lato), pated (but see above regarding Ischnocnesome of the other genera will have to be redemarcated. ma), no synapomorphy has yet been identi­

200 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

fied to unite the species of ‘‘ Eleutherodac­ oventrad between that muscle and the deeper tylus ’’ (sensu stricto) (J.D. Lynch and m. adductor mandibulae posterior (‘‘E’’ mus­ Duellman, 1997; but see below), and the E. culature), either. Instead, V 3 lies entirely posbinotatus group in particular was delimited terior to both muscles and runs ventrolaterad only by overall similarity and biogeographic toward the jaw—that is, it does not run proximity (J.D. Lynch , 1976). around the anterior face of the m. adductor

It should be noted that, although no syn­ mandibulae posterior. J.D. Lynch (1986) reapomorphy is known for ‘‘ Eleutherodacty­ ported a similar pattern for one of three speclus ’’ (sensu stricto), J.D. Lynch and Duell­ imens of ‘‘E.’’ angelicus and one of two man (1997) argued that a large number of its specimens of ‘‘E.’’ maussi (now ‘‘E.’’ biporspecies form a clade delimited by the fifth catus —Savage and Myers, 2002; the other toe being much longer than the third. Insofar specimens exhibited the ‘‘E’’ condition), and as neither of the species of ‘‘ Eleutherodac­ further sampling could show the present obtylus ’’ (sensu stricto) in our sample exhibits servations to be individual anomalies as well. this state, this hypothesis remains to be tested It should also be noted that we have not excritically. amined the m. adductor mandibulae of the

Our results also indicate that Craugastor , other species of the ‘‘E.’’ fraudator group. the so­called Middle American clade delim­ Nevertheless, these observations are reason ited by the synapomorphic ‘‘E’’ pattern of enough to question the placement of this Bothe m. adductor mandibulae (J.D. Lynch , livian group in Craugastor , which is further 1986), was polyphyletic. However, the Mid­ validated by the strongly supported placedle American species we sampled, repre­ ment of ‘‘E.’’ pluvicanorus well outside of senting 5 of the 11 Middle American groups the Craugastor clade. Consequently, we rerecognized by J.D. Lynch (2000)— C. bufon­ move the ‘‘E.’’ fraudator group from Crauiformis, C. bufoniformis group; C. alfredi , C. gastor and return it to the already demonalfredi group; C. augusti , C. augusti group; strably polyphyletic ‘‘ Eleutherodactylus ’’, C. punctariolus and C. cf. ranoides , C. ru­ where J.D. Lynch and McDiarmid (1987) gulosus group; and C. rhodopis , C. rhodopis placed ‘‘E.’’ fraudator originally. Another group—were monophyletic, and the sole out­ option would be to name the ‘‘E.’’ fraudator lier was the Bolivian species ‘‘ Eleuthorodac­ group as a new genus. However, the relatylus’ ’ pluvicanorus . De la Riva and Lynch tionship of this group to ‘‘E.’’ mercedesae (1997) placed this species and ‘‘E.’’ frauda­ (which shares with this group the occurrence tor (grouped subsequently with ‘‘E.’’ ash­ of a frontoparietal fontanelle; J.D. Lynch and kapara as the ‘‘E.’’ fraudator group by Köh­ McDiarmid, 1987) and the hundreds of other ler, 2000) in Craugastor on the basis of its unsampled brachycephalids is unknown, and jaw musculature, although they noted that no given that its placement in ‘‘ Eleutherodacother species of Craugastor is known to ex­ tylus ’’ (sensu stricto) simply inflicts additiontend farther south than northwestern Colom­ al damage on an already polyphyletic genus, bia (e.g., C. bufoniformis ; J.D. Lynch , 1986; we consider it to be premature to name this J.D. Lynch and Duellman, 1997), a possible group at present. but certainly unexpected biogeographic sce­ With the exclusion from nominal Craunario. gastor of the ‘‘E.’’ fraudator group, which

Dissection of the jaw muscles of two spec­ J.D. Lynch (2000) considered to be outside imens of ‘‘E.’’ pluvicanorus (both sides of the scope of his paper, Craugastor corre­ AMNH A165194, right side of AMNH sponds to the clade subtended (appendix 5) A165211) showed it to differ from the ‘‘E’’ our topology to branch 351, and generally pattern of other species (T. Grant, personal corroborates the topology of Craugastor sugobs.). A single muscle (the m. adductor man­ gested by J.D. Lynch (2000). Within Craudibulae externus) originates on the zygomatic gastor, J.D. Lynch (2000: 151, his fig. 9 ) ramus of the squamosal, and the mandibular proposed a clade delimited by extreme sex­

ramus of the trigeminal nerve (V 3) does not ual dimorphism in tympanum size. In our lie lateral (superficial) to it (so it is not the tree C. alfredi , C. augusti , and C. bufonifor­ ‘‘S’’ pattern), but it does not extend poster­ mis, all with nondimorphic tympana, form a

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 201

basal grade, while C. punctariolus , C. rho­ Cryptobatrachidae new family and [368] dopis, and C. cf. ranoides , with strongly sex­ Tinctanura new taxon. ually dimorphic tympana, are monophyletic. CHARACTERIZATION AND DIAGNOSIS: None ‘‘ Pelorius ’’ has had four allozymic fea­ of the morphological characters in our analtures suggested to be synapomorphies ysis (from Haas, 2003) optimize on this (Hedges, 1989), but as noted earlier, this is branch, so its recognition depends entirely on based on sparse taxon sampling. J.D. Lynch molecular evidence, which is decisive. (See (1996) suggested not only that are there no appendix 5 for molecular synapomorphies morphological synapomorphies of this associated with this taxon.) group, but that there is a lot less than meets the eye in Hedges’ (1989) study, particularly [367] FAMILY: CRYPTOBATRACHIDAE NEW

FAMILY with respect to how the allozymic data were interpreted. Alternatively, J.D. Lynch (1996: IMMEDIATELY MORE INCLUSIVE TAXON: 153) suggested that ‘‘ Pelorius ’’ is united [366] Cladophrynia new taxon. with at least some ‘‘ Euhyas ’’ by the posses­ SISTER TAXON: [368] Tinctanura new taxsion of an epiotic flange. So, the evidence for on. ‘‘ Pelorius ’’ and ‘‘ Euhyas ’’ monophyly seems RANGE: Northern Andes and Sierra Santa to be equivocal as well. Marta of Colombia; moderate to high ele­ In summary, we recognize 16 genera with­ vations of the Guayana Shield in Guyana, in Brachycephalidae . Based on our limited Venezuela, and adjacent Brazil. sampling, we recognize as monophyletic CONTENT: Cryptobatrachus Ruthven, 1916 Craugastor , Syrrhophus , Phrynopus , as du­ (type genus of the family); Stefania Rivero , biously monophyletic ‘‘ Euhyas ’’ and ‘‘ Pelor­ 1968 ‘‘1966’’. ius ’’; and as demonstrably nonmonophyletic CHARACTERIZATION AND DIAGNOSIS: Cryp­ ‘‘ Eleutherodactylus ’’ (sensu stricto). We in­ tobatrachidae is characterized by clawcluded in our analysis, but did not test the shaped terminal phalanges and intercalary elmonophyly of Barycholos , Brachycephalus , ements (like Hylidae , Hemiphractidae , and and Ischnocnema , all of which fall within Amphignathodontidae ) and endotrophic lar­ ‘‘ Eleutherodactylus ’’ (sensu stricto). We did vae that develop on the back of the adult not include any representative of Adelophry­ (like Hemiphractidae and Amphignathodonne, Atopophrynus , Dischidodactylus , Eupar­ tidae). Unlike Amphignathodontidae , but like kerella, Geobatrachus , Phyllonastes , and Hemiphractidae , Cryptobatrachidae does not Phyzelaphryne . (See appendix 7 for new develop a dorsal pouch but differs from combinations produced by these generic Hemiphractus in lacking fang­like teeth. changes.) None of the morphological characters in our

analysis (from Haas, 2003) optimize on this

branch, because no member of Cryptoba­

[366] CLADOPHRYNIA NEW TAXON trachidae was studied by Haas (2003). (Mo­ ETYMOLOGY: Clados (Greek: branch) 1 lecular transformations associated with this phrynos (Greek: toad), referring to the ob­ taxon are listed in appendix 5.) servation that this taxon is a clade but not obviously united by any morphological syn­ [368] TINCTANURA NEW TAXON apomorphies. ETYMOLOGY: Tincta (Greek: colored, tint­ IMMEDIATELY MORE INCLUSIVE TAXON: ed) 1 anoura (Greek: frog), denoting the fact [349] Meridianura new taxon. that many of the frogs in this clade are spec­ SISTER TAXON: [350] Brachycephalidae tacularly colored (although some groups Günther, 1858. within it—notably, most species in Bufoni­ RANGE: Coextensive with Anura , exclud­ dae—certainly lack this characteristic). ing New Zealand, Madagascar, and the Sey­ IMMEDIATELY MORE INCLUSIVE TAXON:

chelles. [366] Cladophrynia new taxon. CONCEPT AND CONTENT : Cladophrynia is a SISTER TAXON: [367] Cryptobatrachidae monophyletic taxon composed of [367] new family.

202 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

RANGE: Cosmopolitan in temperate and the group. Amphignathodontidae can be diftropical areas of the continents, Madagascar, ferentiated from other frog taxa by its pos­ Seychelles, and New Zealand. session of a dorsal pouch for brooding eggs, CONCEPT AND CONTENT : Tinctanura is a a likely synapomorphy. Molecular synapomonophyletic taxon containing [369] Am­ morphies are presented in appendix 5.

phignathodontidae Boulenger, 1882, and

[371] Athesphatanura new taxon. [371] ATHESPHATANURA NEW TAXON CHARACTERIZATION AND DIAGNOSIS: None ETYMOLOGY: Athesphatos (Greek: inexof the morphological characters in our anal­ pressible, marvelous) 1 anoura (Greek: tailysis optimized on this branch, so its recog­ less, i.e., frog), denoting the fact that even nition depends entirely on molecular data. though much research has been done on (See appendix 5 for molecular synapomor­ these frogs, they continue to surprise.

phies associated with this taxon.) IMMEDIATELY MORE INCLUSIVE TAXON: [369] FAMILY: AMPHIGNATHODONTIDAE [368] Tinctanura new taxon.

BOULENGER, 1882 SISTER TAXON: [369] Amphignathodontidae Boulenger, 1882 .

Amphignathodontidae Boulenger, 1882: 449 . RANGE: Coextensive with Anura , exclud­ Type genus: Amphignathodon Boulenger, 1882 . ing Madagascar, Seychelles, and New Zea­ Gastrothecinae Noble, 1927: 93. Type genus:

Gastrotheca Fitzinger, 1843 . land.

Opisthodelphyinae Lutz, 1968: 13. Type genus CONCEPT AND CONTENT : As here conceived, Opisthodelphys Günther, 1859 ‘‘1858’’. Athesphatanura is a monophyletic group composed of [372] Hylidae Rafinesque, IMMEDIATELY MORE INCLUSIVE TAXON: 1815, and [424] Leptodactyliformes new [368] Tinctanura new taxon. taxon.

SISTER TAXON: [371] Athesphatanura new CHARACTERIZATION AND DIAGNOSIS: Athestaxon. phatanura is a monophyletic group composed RANGE: Costa Rica and Panama, northern of Hylidae and the bulk of the former non­ and western South America southward to brachycephalid, non­batrachophrynid leptonorthwestern Argentina ; eastern and south­ dactylids. The following characters suggesteastern Brazil; Trinidad and Tobago. ed by Haas (2003) on the basis of a relatively CONTENT: Flectonotus Miranda­Ribeiro , small number of exemplars are potential syn­ 1920 ; Gastrotheca Fitzinger, 1843 . apomorphies of this group: (1) pars alaris and CHARACTERIZATION AND DIAGNOSIS: Haas pars corporis separated by deep distal notch (2003) suggested the following characters (Haas 86.1); (2) commissura proximalis II that optimize on his exemplar Gastrotheca absent ( Haas 110.0) ; and (3) commissura riobambae of amphignathodontids and may proximalis III absent (Haas 111.0). In addibe synapomorphies of Amphignathodonti­ tion, molecular synapomorphies are summadae: (1) m. subarcualis rectus I portion with rized in appendix 5.

origin from ceratobranchial III absent (Haas

35.0); (2) functional larval m. levator man­ [372] FAMILY: HYLIDAE RAFINESQUE, 1815 dibulae lateralis present (Haas 56.0); (3) ra­

mus mandibularis (cranial nerve V 3) poste­ IMMEDIATELY MORE INCLUSIVE TAXON: rior runs through the m. levator mandibulae [371] Athesphatanura new taxon.

externus group (Haas 65.1); (4) posterior pal­ SISTER TAXON: [424] Leptodactyliformes atoquadrate clearly concave with bulging and new taxon.

pronounced margin (Haas 68.1); (5) proces­ RANGE: North and South America, the sus pseudopterygoideus long (Haas 77.2); West Indies, and the Australo­Papuan Re­ and (6) dorsal connection from processus gion; temperate Eurasia, including extreme muscularis to commissura quadrato­orbitalis northern Africa and the Japanese Archipela­ (Haas 78.2). All of these have the potential go.

to be synapomorphies of Amphignathodon­ CONTENT: [386] Hylinae Rafinesque, 1815 tidae, although some or all may be located 1 [373] ([374] Phyllomedusinae Günther , as less inclusive levels of universality within 1858 1 [377] Pelodryadinae Günther, 1858 ).

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 203

CHARACTERIZATION AND DIAGNOSIS: Mor­ don Miranda­Ribeiro, 1920; Aplastodiscus phological characters in our analysis (from Lutz In Lutz, 1950; Argenteohyla Trueb, Haas, 2003 ) that optimize on this branch are 1970; Bokermannohyla Faivovich et al. , (1) m. mandibulolabialis superior present 2005; Bromeliohyla Faivovich et al., 2005 ; (Haas 50.1); and (2) shape of arcus subocu­ Charadrahyla Faivovich et al., 2005 ; Corylaris in cross­section with margin sloping thomantis Boulenger, 1896; Dendropsophus ventrally and laterally (Haas 82.1). As noted Fitzinger, 1843; Duellmanohyla Campbell by Haas (2003), traditional characters of Hy­ and Smith, 1992; Ecnomiohyla Faivovich et lidae, such as claw­shaped terminal phalan­ al., 2005; Exerodonta Brocchi, 1879 ; Hyla ges and intercalary phalangeal elements, do Laurenti, 1768; Hyloscirtus Peters, 1882 ; not optimize as synapomorphies of this Hypsiboas Wagler, 1830 ; Isthmohyla Faigroup because they are shared with Crypto­ vovich, et al., 2005; Itapotihyla Faivovich et batrachidae, Hemiphractidae , and Amphig­ al., 2005; Lysapsus Cope, 1862 ; Megastonathodontidae. matohyla Faivovich et al., 2005; Myersiohyla COMMENT : Because Hylidae is so large, we Faivovich et al., 2005; Nyctimantis Boulendeviate from our practice of providing ac­ ger, 1882; Osteocephalus Steindachner , counts only for families and higher taxa and 1862; Osteopilus Fitzinger, 1843 ; Phyllodyhere provide accounts for the three nominal tes Wagler, 1830; Plectrohyla Brocchi, 1877 ; subfamilies of Hylidae , which have been Pseudacris Fitzinger, 1843 ; Pseudis Wagler , very recently revised (Faivovich et al., 1830; Ptychohyla Taylor, 1944 ; Scarthyla 2005). Duellman and de Sa´, 1988; Scinax Wagler, 1830 ; Smilisca Cope, 1865 (including Pter­ [386] SUBFAMILY: HYLINAE RAFINESQUE, 1815 nohyla Boulenger, 1882); Sphaenorhynchus Hylarinia Rafinesque, 1815: 78 . Type genus: Hy­ Tschudi, 1838; Tepuihyla Ayarzagüena, Señlaria Rafinesque, 1814 . aris, and Gorzula, 1993 ‘‘1992’’; Tlalocohyla Hylina Gray, 1825: 213 . Type genus: Hyla Lau­ Faivovich et al., 2005; Trachycephalus renti, 1768. Tschudi, 1838 (including Phrynohyas Fitzin­ Dryophytae Fitzinger, 1843: 31. Type genus: Dry­ ger, 1843); Triprion Cope, 1866 ; Xenohyla ophytes Fitzinger, 1843 . Izecksohn, 1998 ‘‘1996’’.

Dendropsophi Fitzinger, 1843: 31. Type genus: CHARACTERIZATION AND DIAGNOSIS: Only Dendropsophus Fitzinger, 1843 . one morphological character in our analysis Pseudae Fitzinger, 1843: 33. Type genus: Pseudis optimizes on this taxon: two clearly separate Wagler, 1830.

Acridina Mivart, 1869: 292. Type genus: Acris heads of m. subarcualis obliquus originate Duméril and Bibron, 1841. from ceratobranchialia II and III (character Cophomantina Hoffmann, 1878: 614. Type genus: 31.1 of Haas, 2003). Faivovich et al. (2005) Cophomantis Peters, 1870 . noted another morphological synapomorphy: Lophiohylinae Miranda­Ribeiro, 1926: 64. Type tendo superficialis digiti V (manus) with an genus: Lophyohyla Miranda­Ribeiro, 1926 . additional tendon that arises ventrally from Triprioninae Miranda­Ribeiro, 1926: 64. Type ge­ the m. palmaris longus (da Silva In Duellnus: Triprion Cope, 1866 . man, 2001) and, likely, the 24 chromosome Trachycephalinae Lutz, 1969: 275. Type genus: condition. Nevertheless, substantial numbers Trachycephalus Tschudi, 1838 . of molecular synapomorphies exist (appen­ IMMEDIATELY MORE INCLUSIVE TAXON: dix 5).

[372] Hylidae Rafinesque, 1815 . SYSTEMATIC COMMENTS: The latest revision SISTER TAXON: [373] unnamed taxon of this taxon was by Faivovich et al. (2005), ([374] Phyllomedusinae Günther, 1858 1 who provided a much larger analysis, denser [377] Pelodryadinae Günther, 1858 ). taxonomic sampling, and more molecular RANGE: North and South America, the data per terminal than we do. Our results, West Indies; temperate Eurasia, including ex­ therefore, do not constitute a sufficient test treme northern Africa and the Japanese Ar­ of those results. Nevertheless, differences

chipelago. were noted. We did not find either Hypsiboas CONTENT : Acris Duméril and Bibron , or Hyla to be monophyletic. We presume that 1841; Anotheca Smith, 1939 ; Aparaspheno­ these differences are due to our less­dense

204 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

taxon sampling and application of fewer data monophyly of Pelodryadinae remains unsetthan in the earlier study (Faivovich et al., tled. One character suggested by Haas (2003) 2005). that may optimize on this branch is larval Faivovich et al. (2005) recognized four upper labial papillation complete (Haas 8.0), tribes within Hylinae : (1) Cophomantini which is a reversal from the Phthanobatra­ Hoffmann, 1878 ( Aplastodiscus , Bokerman­ chian condition. However, the number of penohyla, Hyloscirtus , Hypsiboas , and Myer­ lodryadines with complete papillation is siohyla); (2) Dendropsophini Fitzinger, 1843 small, and because Cruziohyla and Phryno­ ( Dendropsophus , Lysapsus , Pseudis , Scar­ medusa (basal taxa in Phyllomedusinae ) also thyla, Scinax , Sphaenorhynchus , and Xeno­ have complete papillation it may be that this hyla); (3) Hylini Rafinesque, 1815 ( Acris , characteristic is a synapomorphy of Phyllo­ Anotheca , Bromeliohyla , Charadrahyla , medusinae 1 Pelodryadinae . Alternatively, Duellmanohyla , Ecnomiohyla , Exerodonta , more dense sampling may show convergence Hyla , Isthmohyla , Megastomatohyla , Plec­ between the phyllomedusinae condition and trohyla, Pseudacris , Ptychohyla , Smilisca , that found in pelodryadines, with this con­ Tlalocohyla , Triprion ); and (4) Lophiohylini dition in pelodryadines, a character of some Miranda­Ribeiro, 1926 ( Aparasphenodon , subset of ‘‘ Litoria ’’ 1 Nyctimystes 1 Cy­ Argenteohyla , Corythomantis , Itapotihyla , clorana.

Nyctimantis , Osteocephalus , Osteopilus, Haas (2003) recovered the subfamily as Phyllodytes , Tepuihyla , and Trachycephal­ paraphyletic with respect to phyllomedusines us). We refer the reader to that revision for on the basis of six exemplars. Tyler (1971c) a detailed discussion of the phylogenetics of noted the presence of supplementary elethe group. ments of the m. intermandibularis in both Pelodryadinae (apical) and Phyllomedusinae [377] SUBFAMILY: PELODRYADINAE (posterolateral). These characters were inter­ GÜNTHER, 1858 preted by Duellman (2001) as nonhomolo­ Pelodryadidae Günther, 1858b: 346 . Type genus: gous and therefore synapomorphies of their Pelodryas Günther, 1858 . respective groups. If these conditions are ho­ Chiroleptina Mivart, 1869: 294. Type genus: Chi­ mologues, however, the polarity between the roleptes Günther, 1858. two states is ambiguous because either, the Cycloraninae Parker, 1940: 12. Type genus: Cy­ pelodryadine or the phyllomedusinae condiclorana Steindachner, 1867. tion, might be ancestral at the Phyllomedu­ Nyctimystinae Laurent, 1975: 183. Type genus: sinae 1 Pelodryadinae level of generality Nyctimystes Stejneger, 1916 . (Faivovich et al., 2005).

IMMEDIATELY MORE INCLUSIVE TAXON: One character in our analysis (originally [373] unnamed taxon composed of [374] from Haas, 2003) optimizes on an [373] un­ Phyllomedusinae Günther, 1858 1 [377] Pe­ named taxon joining Pelodryadinae and lodryadinae Günther, 1858). Phyllomedusinae : ramus mandibularis (cra­ SISTER TAXON: [374] Phyllomedusinae nial nerve V 3) posterior runs through m. le­ Günther, 1858. vator mandibulae externus group (Haas RANGE: Australia and New Guinea; intro­ 65.1). As noted by Faivovich (2005), howduced into New Zealand. ever, another morphological synapomorphy CONTENT: Litoria Tschudi, 1838 (including of Phyllomedusinae 1 Pelodryadinae is the Cyclorana Steindachner, 1867 , and Nycti­ presence of a tendon of the m. flexor ossis mystes Stejneger, 1916; see Systematic Com­ metatarsi II arising only from distal tarsi 2– ments and appendix 7). 3. See also appendix 5 for molecular syna­ CHARACTERIZATION AND DIAGNOSIS: No pomorphies of Phyllomedusinae 1 Pelodrymorphological character optimizes unambig­ adinae.

uously as a synapomorphy of Pelodryadinae . The extensive paraphyly of ‘‘ Litoria ’’ with The molecular data, however, are decisive respect to Cyclorana and ‘‘ Nyctimystes ’’ re­

(see Systematic Comments below and appen­ mains the elephant in the room for Australian dix 5). herpetology, and for reasons that escape us SYSTEMATIC COMMENTS: Evidence of this spectacular problem has largely been ig­

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 205

nored until recently; S. Donnellan and col­ present (Haas 96.2); (5) cleft between hyal laborators are currently addressing pelodry­ arch and branchial arch I closed (Haas adine relationships. A further dimension of 123.0); and (6) pupil shape vertically ellipthis problem is that our results not only reject tical (Haas 143.0). Additionally, Faivovich Litoria monophyly; they also show Nycti­ (2005) noted that ventrolateral position of the mystes nonmonophyly, even though morpho­ spiracle is a likely synapomorphy. logical evidence would suggest that Nyctimystes is monophyletic. Our resolution at [424] LEPTODACTYLIFORMES NEW TAXON this time is to consider Nyctimystes as a syn­ ETYMOLOGY: Leptodactyli­ (with reference onym of Litoria and Cyclorana as a subge­ to the former leptodactylids [Greek: leptos 5 nus within Litoria . It is unfortunate to have narrow 1 dactylos 5 toe]) 1 formes [Greek: to embrace such an uninformative taxonomy, shaped]). but the generic taxonomy as it exists is se­ IMMEDIATELY MORE INCLUSIVE TAXON: riously misleading and no good alternatives [371] Athesphatanura new taxon. present themselves pending the resolution of SISTER TAXON: [372] Hylidae Rafinesque , this problem by S. Donnellan and collabo­ 1815. rators. (See appendix 7 for new combinations RANGE: Coextensive with Anura , excludproduced by these generic changes.) ing Australo­Papuan region, Madagascar,

Seychelles, and New Zealand.

[374] SUBFAMILY: PHYLLOMEDUSINAE CONCEPT AND CONTENT : Leptodactylifor­

GÜNTHER, 1858 mes is a monophyletic taxon composed of Phyllomedusidae Günther, 1858b: 346 . Type ge­ [425] Diphyabatrachia new taxon and [440] nus: Phyllomedusa Wagler, 1830 . Chthonobatrachia new taxon. Pithecopinae Lutz, 1969: 274. Type genus: Pithe­ CHARACTERIZATION AND DIAGNOSIS: Becopus Cope, 1866. yond our molecular data, no characters in our IMMEDIATELY MORE INCLUSIVE TAXON: analysis (originally from Haas, 2003) opti­ [373] unnamed taxon composed of [374] mize on this branch. (See appendix 5 for mo­ Phyllomedusinae Günther, 1858 and [377] lecular synapomorphies of this taxon.) J.D. Pelodryadinae Günther, 1858 . Lynch (1973) reported most of the taxa out­ SISTER TAXON: [377] Pelodryadinae Gün­ side of Leptodactyliformes to have moderther, 1858. ately to broadly dilated sacral diapophyses, RANGE: Tropical Mexico to Argentina. so round sacral diapophyses may be a syna­ CONTENT: Agalychnis Cope, 1864 ; Cru­ pomorphy of this taxon, although reversed in ziohyla Faivovich et al., 2005; Hylomantis Centrolenidae and Bufonidae . Peters, 1873 ‘‘1872’’; Pachymedusa Duell­ [425] DIPHYABATRACHIA NEW TAXON man, 1968; Phasmahyla Cruz, 1991 ‘‘1990’’; Phrynomedusa Miranda­Ribeiro , ETYMOLOGY: Diphya­ (Greek: two­nature) 1923; Phyllomedusa Wagler, 1830 . 1 batrachos (Greek: frog), referencing the CHARACTERIZATION AND DIAGNOSIS: Phyl­ fact that the two components of this taxon lomedusinae is a group of bizarre hylids ( Centrolenidae and Leptodactylidae ) have characterized by vertical pupils and a loris­ very different morphologies and life­histolike movement. Haas (2003) suggested sev­ ries. eral larval characters that are good candi­ IMMEDIATELY MORE INCLUSIVE TAXON: dates for being synapomorphies of Phyllo­ [424] Leptodactyliformes new taxon. medusinae, although they could also be syn­ SISTER TAXON: [440] Chthonobatrachia apomorphies of less inclusive groups: (1) new taxon. suspensorium ultralow (Haas 71.3); (2) pro­ RANGE: Neotropics of North, Central, and cessus pseudopterygoideus short (Haas South America. 77.1); (3) arcus subocularis with three dis­ CONCEPT AND CONTENT : Diphyabatrachia is

tinct processes (Haas 81.2); (4) cartilaginous a monophyletic taxon containing [426] Cenroofing of the cavum cranii with taeniae trolenidae Taylor, 1951, and [430] Leptodactransversalis et medialis (fenestrae parietales) tylidae Werner, 1896 (1838).

206 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

CHARACTERIZATION AND DIAGNOSIS: No on the basis of anatomical and external simmorphological characters suggested by Haas ilarity (Noble, 1931), but subsequent molec­ (2003) optimize on this branch, although ular work (Austin et al., 2002; Faivovich et substantial amounts of molecular evidence al., 2005) has substantiated this relationship. are synapomorphic (see appendix 5). We recognize within Centrolenidae the subfamilies Allophryninae for Allophryne , and [426] FAMILY: CENTROLENIDAE TAYLOR, 1951 [427] Centroleninae , for Centrolene , Cochranella , and Hyalinobatrachium . Centrolen­ Centrolenidae Taylor, 1951: 36 . Type genus: Cen­ inae is united by the possession of intercalary trolene Jiménez de la Espada, 1872.

Allophrynidae Goin et al., 1978: 240 . Type genus: elements between the ultimate and penulti­ Allophryne Gaige, 1926 . New synonym. mate phalanges, fusion of the fibula and tibia (Taylor, 1951; but see Sanchíz and de la IMMEDIATELY MORE INCLUSIVE TAXON: Riva, 1993), and the presence of a medial [425] Diphyabatrachia new taxon. projection on the third metacarpal (Hayes SISTER TAXON: [430] Leptodactylidae Wer­ and Starrett, 1981 ‘‘1980’’).

ner, 1896 (1838). Our results showed ‘‘ Centrolene ’’ to be RANGE: Tropical southern Mexico to Bo­ paraphyletic with respect to ‘‘ Cochranella ’’. livia, northeastern Argentina, and southeast­ Morphological evidence for the monophyly ern Brazil. of Centrolene consists of a single synapo­ CONTENT: Allophryne Gaige, 1926 ; ‘‘ Cen­ morphy, the presence of a humeral spine in trolene ’’ Jiménez de la Espada, 1872 (see adult males (Ruiz­Carranza and Lynch , Systematic Comments); ‘‘ Cochranella ’’ Tay­ 1991a), which is conspicuously present (allor, 1951; Hyalinobatrachium Ruiz­Carran­ beit morphologically different) in both ‘‘ Cenza and Lynch , 1991. trolene ’’ geckoideum and ‘‘ Centrolene ’’ pro­ CHARACTERIZATION AND DIAGNOSIS: Haas soblepon (Ruiz­Carranza and Lynch , 1991b: (2003) suggested several characters for his 3, their fig. 1). Nevertheless, the humerus of exemplar, Cochranella granulosa , that are some species of Cochranella exhibits a concandidates for being synapomorphies of Cen­ spicuously developed ventral crest (e.g., C. trolenidae (including Allophryne , which Haas armata, C. balionota , and C. griffithsi ; J.D. did not examine): (1) anterior insertion of m. Lynch and Ruiz­Carranza, 1997 ‘‘1996’’; see subarcualis rectus II–IV on ceratobranchial also Ruiz­Carranza and Lynch , 1991a), III (Haas 37.2); (2) larval m. levator man­ which at least suggests that coding this chardibulae externus present as one muscle body acter as the presence or absence of a humeral (Haas 54.0); (3) processus anterolateralis of spine may be simplistic. Indeed, the basis is crista parotica absent (Haas 66.0); (4) partes unclear for coding the bladelike ‘‘spine’’ of corpores medially separate (Haas 87.0); (5) Centrolene grandisonae as the same condicleft between hyal arch and branchial arch I tion as the smooth, rounded, and protruding closed (Haas 123.0); and (6) terminal pha­ spine of C. geckoideum and as distinct from langes T­shaped (Haas 156.2). Cochranella the strongly developed bladelike crest of C. granulosa , Haas’ examplar species, lacks lar­ armata. We urge centrolenid workers to exval labial keratodonts (Haas 3.0) but this is amine the different ‘‘spines’’ in greater detail unlikely to be a synapomorphy of the group and to evaluate hypothesized homologies (Altig and McDiarmid, 1999). Burton carefully. (Note that our findings do not rule (1998a) provided evidence suggesting that out homology of the humeral spines. It is the ventral origin of the m. flexor teretes III equally parsimonious for it to have been relative to the corresponding m. transversi gained independently in the two lineages of metacarporum I is a synapomorphy. Regard­ ‘‘ Centrolene ’’ or gained once and lost in less of whether all of these only apply to Cochranella .)

Centroleninae , there are substantial numbers We did not test the monophyly of Cochof molecular synapomorphies (see appendix ranella or Hyalinobatrachium . No synapo­

5). morphy has been identified for Cochranella SYSTEMATIC COMMENTS : The association of (Ruiz­Carranza and Lynch , 1991a) and Darst Allophryne with centrolenids was first made and Cannatella (2004; fig. 22) have presented

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 207

molecular evidence for its nonmonophyly, 1863; Hydrolaetare Gallardo, 1963 ; Leptowhereas Hyalinobatrachium is delimited by dactylus Fitzinger, 1826 (see Systematic the occurrence of a bulbous liver (Ruiz­Car­ Comments and appendix 7 for treatment of ranza and Lynch , 1991a, 1998). subsidiary taxa and synonyms Adenomera Given our topology, the questions sur­ Steindachner, 1867, Lithodytes Fitzinger , rounding the homology of the humeral 1843, and Vanzolinius Heyer, 1974 ); Paraspines, and the lack of evidence for the telmatobius Lutz and Carvalho, 1958; Physmonophyly of Cochranella , we were tempted alaemus Fitzinger, 1826; Pleurodema Tschuto place Cochranella in the synonymy of di, 1838; Pseudopaludicola Miranda­Ribei­ Centrolene . A behavioral synapomorphy for ro, 1926; Scythrophrys Lynch, 1971 ; Somunthe inclusive clade is male–male physical curia Lynch , 1978.

combat undertaken by hanging upside down CHARACTERIZATION AND DIAGNOSIS: This by the feet and grappling venter­to­venter taxon corresponds reasonably closely to the (Bolívar­G. et al., 1999), a behavior other­ former Leptodactylinae , excluding Limnowise known only in phyllomedusines (Py­ medusa (to Cycloramphidae ) and adding burn, 1970; Lescure et al., 1995; Wogel et Paratelmatobius and Scythrophrys (from the al., 2004) and some species of Hypsiboas and former Cycloramphinae ). Most species are Dendropsophus ( Hylidae ; J. Faivovich and found on the forest floor, although a diversity C.F.B. Haddad, personal obs.). However, the of tropical biomes are inhabited. Many speresulting genus, though monophyletic, would cies are foam­nest builders (excluding Parbe unwieldy (with 100 species). In light of atelmatobius, some Pleurodema, Pseudopathe poverty of our taxon sampling, and our ludicola, Scythrophrys , and Somuncuria ; anticipation of more thorough phylogenetic Barrio, 1977; Pombal and Haddad, 1999; C. studies of this charismatic group, we retain Haddad, personal obs.), and this may be synthe current taxonomy and place quotation apomorphic of the group. Several of the marks around ‘‘ Centrolene ’’ to denote its ap­ characters in our analysis (from Haas, 2003) parent paraphyly and around ‘‘ Cochranella ’’ optimize on our topology on the [431] to denote its nonmonophyly as well. branch subtending Physalaemus and Pleurodema . Because Haas did not study other [430] FAMILY: LEPTODACTYLIDAE WERNER , members of our Leptodactylidae , these char­ 1896 (1838) acters are candidates for being synapomor­ Cystignathi Tschudi, 1838: 26, 78. Type genus: phies of our Leptodactylidae : (1) m. subar­ Cystignathus Wagler, 1830 . cualis rectus I portion with origin from cer­ Leiuperina Bonaparte, 1850: 1 p. Type genus: atobranchial III absent (Haas 35.0); and (2) Leiuperus Duméril and Bibron, 1841 . dorsal connection from processus muscularis Plectromantidae Mivart, 1869: 291. Type genus: to neurocranium and pointed (Haas 78.1).

Plectromantis Peters, 1862 . We also suggest that the bony sternum of Adenomeridae Hoffmann, 1878: 613. Type genus: the former Leptodactylinae (J.D. Lynch , Adenomera Steindachner, 1867 . 1971) is a synapomorphy of this taxon, but Leptodactylidae Werner, 1896: 357 . Type genus: reversed to the cartilaginous condition in the Leptodactylus Fitzinger, 1826 .

[435] branch subtending Paratelmatobius 1 Pseudopaludicolinae Gallardo, 1965: 84. Type ge­

nus: Pseudopaludicola Miranda­Ribeiro, 1926 . Scythrophrys . The bony sternum occurs independently in Limnomedusa (J.D. Lynch , IMMEDIATELY MORE INCLUSIVE TAXON: 1971) in Cycloramphidae , and a calcified [425] Diphyabatrachia new taxon. sternum occurs in Barycholos (Brachyce­ SISTER TAXON: [426] Centrolenidae Taylor , phalidae; J.D. Lynch , 1980).

1951. SYSTEMATIC COMMENTS: Hydrolaetare Gal­ RANGE: Extreme southern USA and trop­ lardo, 1963, is associated with this group beical Mexico throughout Central America and cause of its presumed association with Lep­ South America. todactylus (Heyer, 1970), although we sug­ the

CONTENT: Edalorhina Jiménez de la Es­ gest that this proposition needs to be evalupada, 1871 ‘‘1870’’; Engystomops Jiménez ated carefully. We place Somuncuria de la Espada, 1872 ; Eupemphix Steindachner , provisionally in this group on the basis of 208 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

evidence (though not the conclusions) sug­ Ceratophryidae Tschudi, 1838 , and [448] gested by J.D. Lynch (1978b), who placed Hesticobatrachia new taxon. Somuncuria as the sister taxon of Pleurode­ CHARACTERIZATION AND DIAGNOSIS: None ma. On the basis of our evidence, Leptodac­ of the larval characters studied by Haas tylus is paraphyletic with respect to Vanzo­ (2003) optimize on this branch, so the diaglinius; this agrees with the results of Heyer nosis of this taxon rests entirely on molecular (1998) who presented evidence to place Van­ evidence. See appendix 5 for molecular synzolinius deeply within a paraphyletic Lepto­ apomorphies. dactylus , and likely the sister taxon of Leptodactylus diedrus . De Sá et al. (2005) also [441] FAMILY: CERATOPHRYIDAE TSCHUDI,

1838 came to this conclusion and placed Vanzolinus in the synonymy of Leptodactylus . We Ceratophrydes Tschudi, 1838: 26. Type genus: regard Vanzolinius as a subjective junior syn­ Ceratophrys Wied­Neuwied, 1824 . onym of Leptodactylus . Although our data Telmatobii Fitzinger, 1843: 31. Type genus: Telare agnostic on the subject, Heyer (1998) and matobius Wiegmann, 1834. New synonym.

Stombinae Gallardo, 1965: 82. Type genus: Stom­ Kokubum and Giaretta (2005) also presented bus Gravenhorst, 1825. evidence that recognizing Adenomera ren­ Batrachylinae Gallardo, 1965: 83. Type genus: ders Leptodactylus paraphyletic and that Batrachylus Bell, 1843. New synonym. Lithodytes is the sister taxon of Adenomera . On the basis of this evidence, as well as Hey­ IMMEDIATELY MORE INCLUSIVE TAXON: er’s (1998) and Kokubum and Giaretta’s [440] Chthonobatrachia new taxon. (2005) evidence, we place Adenomera Stein­ SISTER TAXON: [448] Hesticobatrachia new dachner, 1867, as a synonym of Lithodytes taxon. Fitzinger, 1843, and Lithodytes as a subgenus RANGE: Southern Andean and tropical

lowland South America from Colombia and of Leptodactylus , without delimiting any oth­ Venezuela south to extreme southern Argen­ er subgenera so as not to construct or imply tina and Chile. any paraphyletic groups (see appendix 7 for CONTENT: Atelognathus Lynch, 1978 ; Banew combinations). Leptodactylus , therefore trachyla Bell, 1843; Ceratophrys Wied­Neuis equivalent to the taxon subtended by wied, 1824; Chacophrys Reig and Limeses , branch 436 in our tree. 1963; Insuetophrynus Barrio, 1970 (see Sys­ J.D. Lynch (1971: 26) noted that Pseudo­ tematic Comments); Lepidobatrachus Budgpaludicola and Physalaemus (including En­ ett, 1899; Telmatobius Wiegmann, 1834 . gystomops and Eupemphix in his sense) share CHARACTERIZATION AND DIAGNOSIS: Morthe feature of dextral vents in larvae, as do phological characters for this group were de­ Edalorhina and some Paratelmatobius (Altig rived solely from Ceratophrys and Lepidob­ and McDiarmid, 1999). atrachus. Inasmuch as these are very derived

taxa, the characters in our analysis that op­ [440] CHTHONOBATRACHIA NEW TAXON timize on them are likely characters of Lep­

idobatrachus 1 Ceratophrys , although some ETYMOLOGY: Chthonos­ (Greek: ground) of them may optimize at other hierarchic lev­ 1 batrachos (Greek: frog), referencing the els (including Ceratophryidae in our sense) fact that most of the included species are once relevant specimens have been examground­dwelling. ined. Relevant morphological characters IMMEDIATELY MORE INCLUSIVE TAXON: (from Haas, 2003) are (1) m. diaphragmato­ [424] Leptodactyliformes new taxon. praecordialis absent (Haas 25.0; a reversal SISTER TAXON: [425] Diphyabatrachia new from the phthanobatrachian condition, also in taxon. bufonids, microhylids, and some ranoids); RANGE: Coextensive with Anura , exclud­ (3) mm. levatores arcuum branchialium I and ing Australo­Papuan region, Madagascar, II narrow with a wide gap between them

Seychelles, and New Zealand. (Haas 40.0; a reversal from the phthanoba­ CONCEPT AND CONTENT : Chthonobatrachia trachian condition, also in some Litoria and is a monophyletic group composed of [441] Atelopus ); (4) m. suspensoriohyoideus absent

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 209

(Haas 45.0; a reversal of the acosmanuran We did not study Insuetophrynus and it is condition, also in some hylines and Atelo­ therefore only provisionally allocated to this pus); (5) ramus mandibularis (cranial nerve family. Lynch (1978bb), on the basis of a V 3) runs through the m. levator mandibulae phylogenetic analysis of morphology, consisexternus group (Haas 65.1; one of many re­ tently recovered Insuetophrynus as the sister versals on the overall tree); (6) anterolateral taxon of Atelognathus , while Diaz et al. base of processus muscularis without con­ (1983) considered the relationships of Insuespicuous projection (Haas 86.0); (7) tectum tophrynus to lie with Alsodes (Cycloramphiof cavum cranii almost completely chondri­ dae) or Telmatobius ( Ceratophryidae ). The fied (Haas 96.4); (8) spicula short or absent characters suggested by Diaz et al. (1983) in (Haas 112.0); and (9) branchial food traps support of their arrangement all are likely absent (Haas 134.0). Molecular synapomor­ plesiomorphies, however, so we retain the phies for Ceratophryidae appear in appendix hypothesis of Lynch (1978bb) pending ad­ 5.

ditional evidence.

SYSTEMATIC COMMENTS: Within Ceratophryidae , the association of the genera is relatively weak, with the exception of Lepidob­ [448] HESTICOBATRACHIA NEW TAXON atrachus 1 Ceratophrys 1 Chacophrys and Batrachyla 1 Atelognathus . ETYMOLOGY: Hestico­ (Greek: agreeable,

We recognize two subfamilies within Cer­ pleasing) 1 batrachos (Greek: frog), denotatophryidae: [442] Telmatobiinae (for Tel­ ing the agreeable nature of these frogs, parmatobius) and [444] Ceratophryinae . Within ticularly with respect to the nature of the type Ceratophryinae we recognize two tribes: genus of their sister taxon, Ceratophryidae . [445] Batrachylini (for Atelognathus , Batra­ IMMEDIATELY MORE INCLUSIVE TAXON: chyla, and, presumably Insuetophrynus ), and [440] Chthonobatrachia new taxon. [446] Ceratophryini (for Lepidobatrachus, SISTER TAXON: [441] Ceratophryidae Ceratophrys , and Chacophrys ). Ceratophryi­ Tschudi, 1838. nae has a continuous row of papilla on the RANGE: Coextensive with Anura , excludupper lip in larvae (Haas 8.0), a synapomor­ ing Australo­Papuan region, Madagascar, phy. Batrachylini is also diagnosed on the ba­ Seychelles, and New Zealand. sis of molecular evidence (appendix 5), and CONCEPT AND CONTENT : Hesticobatrachia is Ceratophryini is diagnosed on the basis of a monophyletic group composed of [449] molecular evidence as well as on traditional Cycloramphidae Bonaparte, 1858 , and [460] morphological characters associated with this Agastorophrynia new taxon. cluster of genera (J.D. Lynch , 1971, 1982b):

C: Char­

HARACTERIZATION AND DIAGNOSIS (1) transverse processes of anterior presacral

acters proposed by Haas (2003) that optimize vertebrae widely expanded; (2) cranial bones

on this branch are (1) posterior palatoquaddermosed; (3) teeth fang­like, nonpedicellate;

rate curvature clearly concave with bulging and (4) absence of pars palatina of maxilla

and pronounced margin (Haas 68.1); and (2) and premaxilla. Another character, presence of a vertebral shield, may be a synapomor­

presence of a dorsal connection from procesphy of Ceratophryini although the optimisus muscularis to neurocranium ligament zation of this feature is ambiguous, requires (Haas 78.1). Molecular synapomorphies are detailed study, and was not considered a syn­ summarized in appendix 5. apomorphy by Lynch (1982b). The shield is SYSTEMATIC COMMENTS: We did not study present in Ceratophrys aurita , C. cranwelli , Rupirana Heyer, 1999 , and cannot allocate it, C. ornata , C. joazeirensis , and in Lepidoba­ even provisionally, although Heyer (1999) trachus asper and L. llanensis (in these two thought that it might have some kind of rethe morphology of the shield is quite differ­ lationship, not close, with either Batrachyla ent from that of Ceratophrys ; J. Faivovich, ( Cycloramphidae ) or Thoropa ( Thoropidae ). repersonal obs.), and absent in C. calcarata , C. The data to support either contention are amcornuta, C. testudo , C. stolzmanni , Chaco­ biguous at best. The position of Rupirana phrys pierottii , and Lepidobatrachus laevis . mains to be elucidated.

210 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

[449] FAMILY: CYCLORAMPHIDAE BONAPARTE , molecular synapomorphies (see appendix 5),

1850 Hylodinae is diagnosed by the synapomor­ Cyclorhamphina Bonaparte, 1850: 1 p. Type ge­ phy of having a lateral vector to the alary nus: Cycloramphus Tschudi, 1838 . processes (J.D. Lynch , 1971: 39), T­shaped Rhinodermina Bonaparte, 1850: 1 p. Type genus: terminal phalanges, and dermal scutes on the Rhinoderma Duméril and Bibron, 1841 . New top of the digital discs (J.D. Lynch , 1971, synonym, considered a junior synonym of Cy­ 1973). This latter character is also found in clorhamphina Bonaparte, 1850, under Article Petropedetidae (Ranoides) and Dendrobati­ 24.2.1 (Rule of First Revisor) of the International Code of Zoological Nomenclature dae. (ICZN, 1999). Cycloramphinae is not readily diagnosed Hylodinae Günther, 1858b: 346. Type genus: Hy­ on the basis of morphology, but it is comlodes Fitzinger, 1826. New synonym. posed of two tribes. The first of these is [453] Alsodina Mivart, 1869: 290. Type genus: Alsodes Cycloramphini Bonaparte, 1850 (Cycloram­ Bell, 1843. New synonym. phus, Crossodactylodes , and Zachaenus ), Grypiscina Mivart, 1869: 295. Type genus: Gry­ corresponding to Grypiscini Mivart, 1869, of piscus Cope, 1867 ‘‘1866’’. J.D. Lynch (1971) with the addition of Rhi­ Elosiidae Miranda­Ribeiro, 1923: 827. Type ge­ noderma. The second is [454] Alsodini Minus: Elosia Tschudi, 1838 . Odontophrynini J.D. Lynch , 1969: 3. Type genus: vart, 1869 (composed of the remaining gen­ Odontophrynus Reinhardt and Lütken, 1862 era). Alsodini is diagnosed by its possession ‘‘1861’’. (Odontophrynini subsequently named of Type II cotylar arrangement (cervical comore formally by J.D. Lynch , 1971: 142.) tyles narrowly separated with two distinct ar­

ticular surfaces; J.D. Lynch , 1971). This oc­ IMMEDIATELY MORE INCLUSIVE TAXON: curs otherwise in Hyloides only in Batrach­ [448] Hesticobatrachia new taxon. ophrynidae, Limnodynastidae , Megaelosia, SISTER TAXON: [460] Agastorophrynia new and Telmatobius (J.D. Lynch, 1971) , so is taxon. likely synapomorphic at this level. (See ap­ RANGE: Southern tropical and temperate pendix 5 for relevant molecular synapomor­ South America. phies.) CONTENT: Alsodes Bell, 1843 ; Crossodactylodes Cochran, 1938 ; Crossodactylus Duméril and Bibron, 1841 ; Cycloramphus [460] AGASTOROPHRYNIA NEW TAXON Tschudi, 1838; Eupsophus Fitzinger, 1843 ; Hylodes Fitzinger, 1826 ; Hylorina Bell , ETYMOLOGY: Agastoro­ (Greek: near kins­

man) 1 phrynia (Greek: having the nature of 1843; Limnomedusa Fitzinger, 1843 ; Macro­ a toad), noting the surprisingly close relagenioglottus Carvalho, 1946; Megaelosia Miranda­Ribeiro, 1923 ; Odontophrynus tionship of Dendrobatoidea and Bufonidae . Reinhardt and Lütken, 1862 ‘‘1861’’; Pro­ IMMEDIATELY MORE INCLUSIVE TAXON: ceratophrys Miranda­Ribeiro, 1920 ; Rhinod­ [448] Hesticobatrachia new taxon. erma Duméril and Bibron, 1841; Zachaenus SISTER TAXON: [449] Cycloramphidae Bon­ Cope, 1866. aparte, 1850. CHARACTERIZATION AND DIAGNOSIS: One of RANGE: Coextensive with Anura , excludthe morphological characters suggested by ing Australo­Papuan region, Madagascar, Haas (2003) optimizes as a synapomorphy of Seychelles, and New Zealand. this group: anterior insertion of m. subarcu­ CONCEPT : Agastorophrynia is a monophyalis rectus II–IV on ceratobranchial I (Haas letic taxon composed of [461] Dendrobato­ 37.0). All decisive evidence for the existence idea Cope, 1865, and [469] Bufonidae Gray , of this clade is molecular (see appendix 5). 1825. SYSTEMATIC COMMENTS: Within Cyclor­ CHARACTERIZATION AND DIAGNOSIS: None amphidae we recognize two sister subfami­ of the morphological characters suggested by lies, [450] Hylodinae Günther, 1858 (con­ Haas (2003) optimize in a way that would

taining Crossodactylus , Megaelosia , and Hy­ suggest their possible candidacy as synapolodes) and [452] Cycloramphinae Bonaparte , morphies of Agastorophrynia. All decisive 1850 (for the remaining genera). Other than evidence for the recognition of this taxa is

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 211

molecular. These molecular synapomorphies of paired dermal scutes atop the digits has are summarized in appendix 5. been claimed as a synapomorphy of Dendro­

batidae 1 Hylodinae (e.g, Noble, 1926), but [461] SUPERFAMILY: DENDROBATOIDEA COPE, its optimization on our optimal topology re­

1865 quires only a single extra step, versus the 39 IMMEDIATELY MORE INCLUSIVE TAXON: steps required to disrupt the relationship be­ [460] Agastorophrynia new taxon. tween Thoropa and Dendrobatidae . Insofar SISTER TAXON: [469] Bufonidae Gray , as there is no compelling evidence against 1825. our optimal solution, and despite our aston­ RANGE: Central America ( Nicaragua to ishment at the result, we recognize these sis­ Panama) and South America (Guianas, Am­ ter taxa as Dendrobatoidea and leave it to azon drainage, south to Bolivia and south­ future tests based on greater character (ineastern Brazil). cluding morphology) and taxon sampling to CONTENT: Thoropidae new family and assess the reality of this clade. Alternatively, [462] Dendrobatidae Cope, 1865 . we could have left Thoropa insertae sedis— CHARACTERIZATION AND DIAGNOSIS: Evi­ an obviously deficient solution—or have dence for Dendrobatoidea is derived entirely placed it inside Dendrobatidae . from DNA sequence data, as summarized in appendix 5. FAMILY: THOROPIDAE NEW FAMILY SYSTEMATIC COMMENT : The sister group re­ IMMEDIATELY MORE INCLUSIVE TAXON: lationship of Dendrobatidae and Thoropa , to [461] Dendrobatoidea. our knowledge, has never been proposed, SISTER TAXON: [462] Dendrobatidae Cope , and this is one of the most heterodox of our 1865. results. No morphological synapomorphies RANGE: Eastern, southeastern, and southare apparent, and a large number of charac­ ern Brazil. ters differ between the two taxa. Neverthe­ CONTENT: Thoropa Cope, 1865 . less, evidence for alternative placement of CHARACTERIZATION AND DIAGNOSIS: Be­ Thoropa appears to be lacking (although the cause this taxon was not studied by Haas larvae of Thoropa and Cycloramphus are (2003) none of the morphological characters very similar and semiterrestrial; Haddad and in our analysis could optimize on this branch. Prado, 2005), and most of the characters that All evidence for the phylogenetic placement differ between Dendrobatidae and Thoropa of this taxon as distinct from Cycloramphiare either of unclear polarity or unique to nae is molecular, although Thoropa larvae Dendrobatidae among hyloids (e.g., thigh can be distinguished from all near relatives musculature, epicoracoid fusion and nonov­ by being very attenuate and flattened (J.D. erlap). Furthermore, it does not appear that Lynch , 1971: 124). For additional differentia this result is due to inadequate algorithmic see J.D. Lynch (1972a). searching. At numerous points in the analysis SYSTEMATIC COMMENTS: See comment unwe placed Dendrobatidae , Thoropa , the hy­ der Hesticobatrachia regarding Rupirana . lodine genera, and various other cycloram­ J.D. Lynch (1971) considered Thoropa to be phines and bufonids in alternative arrange­ closely related to Batrachyla , sharing a Type ments and submitted those topologies either I cotylar arrangement, although the polarity as starting points or as constraint files for fur­ of the character was unclear in his study, and ther searching, but our analysis invariably led dendrobatids have the Type I condition as away from those solutions. The Bremer val­ well, rendering this character uninformative. ues and jackknife frequencies are both strong for this clade (39% and 100%, respectively). [462] FAMILY: DENDROBATIDAE COPE, 1865 The arrangement Thoropa (Hylodinae 1 (1850) Dendrobatidae ) requires 56 extra steps, and Phyllobatae Fitzinger, 1843: 32. Type genus: placing Thoropa in the more conventional ar­ Phyllobates Duméril and Bibron, 1841 . :: rangement Thoropa 1 ( Hylorina 1 ( Alsodes Eubaphidae Bonaparte, 1850: 1 p. Type genus 1 Eupsophus ) and (Hylodinae 1 Dendroba­ Eubaphus Bonaparte, 1831 .

tidae) requires 87 extra steps. The occurrence Hylaplesidae Günther, 1858b: 345. Type genus 212 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

Hylaplesia Boie, 1827 (5 Hysaplesia Boie , rently in a state of flux. Dendrobatid mono­ 1826). phyly has been upheld consistently (e.g., Dendrobatidae Cope, 1865: 100 . Type genus: Myers and Ford, 1986; Ford, 1993; Haas, Dendrobates Wagler, 1830 . 2003; Vences et al., 2003b) since first pro­ Colostethidae Cope, 1867: 191 . Type genus: Co­ posed by Noble (1926; see Grant et al., lostethus Cope, 1866.

Calostethina Mivart, 1869: 293. Type genus: Ca­ 1997), but the relationships among dendrolostethus Mivart, 1869. batids remain largely unresolved.

The most generally accepted view of den­ IMMEDIATELY MORE INCLUSIVE TAXON: drobatid systematics, as summarized by My­ [461] Dendrobatoidea Cope, 1865. ers et al. (1991; see also Kaplan, 1997), al­ SISTER TAXON: Thoropidae new family. locates approximately two­thirds of the spe­ RANGE: Central America ( Nicaragua to cies to a ‘‘basal’’ grade of usually dully col­ Panama) and South America (Guianas, Am­ ored, nontoxic frogs (including Aromobates , azon drainage, south to Bolivia and central, Colostethus , Mannophryne , and Nephelobasouthern, and southeastern Brazil). tes), while the remaining one­third is hypoth­ CONTENT: Allobates Zimmermann and esized to form a clade of putatively apose­ Zimmermann, 1988; Ameerega Bauer, 1986 matic frogs (including Allobates , Ameere­ (including Epipedobates Myers, 1987 ); Aro­ ga 30, Dendrobates , Minyobates , Phobobates , mobates Myers, Paolillo O., and Daly, 1991; and Phyllobates ).

Colostethus Cope, 1866 ; Cryptophyllobates Compelling evidence for the monophyly Lötters, Jungfer, and Widmer, 2000; Dendro­ of most genera is lacking. This is especially bates Wagler, 1830 (including Oophaga the case for the ‘‘basal’’ taxa. The nonmon­ Bauer, 1988, and Ranitomeya Bauer, 1986 ); ophyly of Colostethus has been recognized Mannophryne La Marca, 1992 ; Minyobates for decades (J.D. Lynch , 1982a; J.D. Lynch Myers, 1987; Nephelobates La Marca, 1994 ; and Ruiz­Carranza, 1982), and the naming of Phobobates Zimmermann and Zimmermann , Aromobates , Epipedobates , Mannophryne , 1988; Phyllobates Duméril and Bibron , and Nephelobates has merely exacerbated the 1841. problem (Kaiser et al., 1994; Coloma, 1995; CHARACTERIZATION AND DIAGNOSIS: Den­ Meinhardt and Parmalee, 1996; Grant et al., drobatids are well­known, mostly diurnal, 1997; Grant, 1998; Grant and Castro­Herreterrestrial, and frequently brightly colored ra, 1998). Molecular evidence for the monofrogs that have the exotic parental behavior phyly of Mannophryne and Nephelobates of carrying tadpoles on their back to water. was presented by La Marca et al. (2002) and Likely synapomorphies of Dendrobatidae (as Vences et al. (2003b), but the relationships optimized on our topology) from those mor­ of those genera to other dendrobatids are unphological characters reported by Haas clear. Aromobates has been hypothesized to (2003) are (1) insertion of m. rectus cervicis be the monotypic sister group of all other on proximal ceratobranchialia III and IV dendrobatids (Myers et al., 1991), but syna­ (Haas 39.2); (2) adrostral cartilage present pomorphies shared with Mannophryne and but small (Haas 90.1); (3) cartilaginous roof­ Nephelobates , also from the northern Andes, ing of the cavum cranii formed by taeniae cast doubt on that claim. No molecular evitecti medialis only (Haas 96.5); (4) larvae dence has been presented for this taxon.

picked up at oviposition site and transported Among the ‘‘aposematic’’ taxa, only Phylto body of water adhering to dorsum of adult lobates is strongly corroborated as monophy­ (Haas 137.1); (5) amplectic position cephalic letic (Myers et al., 1978; Myers, 1987; (Haas 139.2); (6) guiding behavior (Haas

142.1); (7) firmisterny (Haas 144.1); and (8) 30 As noted earlier, our recognition of Ameerega terminal phalanges T­shaped (Haas 156.2). Bauer, 1986, as a senior synonym of Epipedobates My­ Some of these characters may ultimately ers, 1987, follows the recommendation of Walls (1994). be found to be synapomorphies of Dendro­ Ranitomeya Bauer, 1988 , and Oophaga Bauer, 1994 , are

nomenclaturally valid names, but insofar as they have

batoidea, because Thoropa has not been eval­

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 213

Clough and Summers, 2000; Vences et al., Atelopoda Fitzinger, 1843: 32. Type genus: Ate­ 2000b; Widmer et al., 2000). No synapo­ lopus Duméril and Bibron, 1841 . morphy is known for Ameerega , and it is Phryniscidae Günther, 1858b: 346 . Type genus: likely paraphyletic or polyphyletic with re­ Phryniscus Wiegmann, 1834. Adenomidae Cope, 1861 ‘‘1860’’: 371. Type gespect to Allobates , Colostethus , Cryptophyl­ nus: Adenomus Cope, 1861 . lobates, and Phobobates . Schulte (1989) and Dendrophryniscina Jiménez de la Espada, 1871 Myers et al. (1991) rejected Allobates and ‘‘1870’’: 65. Type genus: Dendrophryniscus Phobobates on the basis of errors in the anal­ Jiménez de la Espada, 1871 ‘‘1870’’. ysis of behavior, lack of evidence, unac­ Platosphinae Fejérváry, 1917: 147. Type genus: counted character conflict, incorrect charac­ Platosphus d’Isle, 1877 (fossil taxon considered ter coding, and creation of paraphyly in to be in this synonymy because Platosophus 5 Ameerega (as also found by Clough and Bufo [sensu lato]). Summers, 2000; Vences et al., 2000b; Santos Bufavidae Fejérváry, 1920: 30. Type genus: Bu­ et al., 2003; Vences et al., 2003b), but many favus Portis, 1885 (fossil taxon considered to be in this synonymy because Bufavus 5 Bufo authors continue to recognize them. In ad­ [sensu lato]). dition, Phobobates was found to be mono­ Tornierobatidae Miranda­Ribeiro, 1926: 19. Type phyletic by Vences et al. (2000b) but para­ genus: Tornierobates Miranda­Ribeiro, 1926 . phyletic by Clough and Summers (2000). Nectophrynidae Laurent, 1942: 6. Type genus: Similarly, Minyobates may or may not be Nectophryne Buchholz and Peters, 1875 . nested within Dendrobates (Silverstone, Stephopaedini Dubois, 1987 ‘‘1985’’: 27. Type 1975; Myers, 1982, 1987; Myers and Bur­ genus: Stephopaedes Channing, 1978 . rowes, 1987; Jungfer et al., 1996; Clough and Summers, 2000; Jungfer et al., 2000). IMMEDIATELY MORE INCLUSIVE TAXON: Likewise, although neither study recognized [460] Agastorophrynia new taxon. Minyobates , it was found to be monophyletic SISTER TAXON: [461] Dendrobatoidea by Santos et al. (2003) but polyphyletic by Cope, 1865. Vences et al. (2003b). Cryptophyllobates is RANGE: Cosmopolitan in temperate and the most recently named genus, but it is tropical areas except for the Australo­Papuan monotypic, and its relationship to other den­ region, Madagascan, Seychelles, and New drobatids is unclear. Zealand. Difficulties in understanding the phyloge­ CONTENT: Adenomus Cope, 1861 ‘‘1860’’; ny of dendrobatid frogs are compounded by Altiphrynoides Dubois, 1987 ‘‘1986’’ (inthe taxonomic problems that surround many cluding Spinophrynoides Dubois, 1987 nominal species and under appreciation of ‘‘1986’’; see Systematic Comments); Amiespecies diversity (Grant and Rodriguez, tophrynus new genus (see Systematic Com­ 2001). Sixty­nine valid species were named ments and appendix 7); Anaxyrus Tschudi , over the past decade (more species than were 1845 (see Systematic Comments and appenknown in 1960), 55 of which were referred dix 7); Andinophryne Hoogmoed, 1985 ; Anto Colostethus . Many nominal species sonia Stoliczka, 1870; Atelophryniscus throughout Dendrobatidae are likely com­ McCranie, Wilson, and Williams, 1989; Ateposed of multiple cryptic species awaiting lopus Duméril and Bibron, 1841 ; Bufo Laudiagnosis (e.g., Caldwell and Myers, 1990; renti, 1768 (see Systematic Comments and Grant and Rodriguez, 2001; Grant, 2002), appendix 7); Bufoides Pillai and Yazdani , but the rapid increase in recognized diversity 1973; Capensibufo Grandison, 1980 ; is not unaccompanied by error, and critical Chaunus Wagler, 1828 (see Systematic evaluation of the limits of nominal taxa will Comments and appendix 7); Churamiti undoubtedly result in some number of these Channing and Stanley, 2002; Cranopsis being placed in synonymy (e.g., Coloma, Cope, 1875 ‘‘1876’’ (see Systematic Com­ 1995; Grant, 2004). ments and appendix 7); Crepidophryne Cope, 1889 ; Dendrophryniscus Jiménez de

[469] FAMILY: BUFONIDAE GRAY, 1825 la Espada, 1871 ‘‘1870’’; Didynamipus An­ Bufonina Gray, 1825: 214. Type genus: Bufo Lau­ dersson, 1903; Duttaphrynus new genus (see renti, 1768. appendix 7); Epidalea Cope, 1865 (see Sys­

214 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

tematic Comments and appendix 7); Frostius unexamined by us; Graybeal and Cannatella, Cannatella, 1986; Ingerophrynus new ge­ 1995) lacks a Bidder’s organ and because our nus; Laurentophryne Tihen, 1960 ; Lepto­ molecular data place Melanophryniscus firmphryne Fitzinger, 1843; Melanophryniscus ly as the sister taxon of remaining bufonids, Gallardo, 1961; Mertensophryne Tihen , the presence of a Bidder’s organ is a syna­ 1960 (including Stephopaedes Channing , pomorphy not of Bufonidae , but of branch 1979 ‘‘1978’’; see Systematic Comments and 470, Bufonidae excluding Melanophryniscus appendix 7); Metaphryniscus Señaris , Ayar­ (and presumably Truebella ). Larval characzagüena, and Gorzula, 1994; Nannophryne ters (from Haas, 2003) that are synapomor­ Günther, 1870 (see Systematic Comments phies of bufonids excluding Melanophrynis­ and appendix 7); Nectophryne Buchholz and cus (and possibly Truebella ) are (1) ramus Peters, 1875; ‘‘ Nectophrynoides ’’ Noble, mandibularis (cranial nerve V 3) runs through 1926 (see Systematic Comment ); Nimba­ the m. levator mandibulae externus group phrynoides Dubois, 1987 ‘‘1986’’; Oreo­ (Haas 65.1); (2) dorsal connection from prophrynella Boulenger, 1895; Osornophryne cessus muscularis to commissura quadrato­ Ruiz­Carranza and Hernández­Camacho, orbitalis (Haas 78.2); (3) eggs deposited in 1976; Parapelophryne Fei, Ye, and Jiang , strings (Haas 141.1, diversely modified high­ 2003; Pedostibes Günther, 1876 ‘‘1875’’; Pe­ er up in the bufonid tree). Atlantal cotyles lophryne Barbour, 1938; Peltophryne Fitzin­ juxtaposed (J.D. Lynch , 1971, 1973) is also ger, 1843; Phrynoidis Fitzinger, 1843 (see a likely synapomorphy of this taxon.

Systematic Comments and appendix 7); SYSTEMATIC COMMENTS: As evidenced by Poyntonophrynus new genus (see Systematic our results Bufo is wildly paraphyletic with Comments and appendix 7); Pseudobufo a number of other nominal genera (as docu­ Tschudi, 1838; Pseudepidalea new genus mented by Graybeal, 1997). Our sampling (see appendix 7); Rhaebo Cope, 1862 (see has likely hardly scratched the surface of this Systematic Comments and appendix 7); problem, and we hope that subsequent work Rhamphophryne Trueb, 1971 ; Rhinella Fit­ will continue to add to the evidence so far zinger, 1826 (see Systematic Comments and presented so that a more universal resolution appendix 7); Schismaderma Smith, 1849 ; may be reached. A complete remedy of the Truebella Graybeal and Cannatella, 1995 ; polyphyly/paraphyly of Bufo is beyond the Vandijkophrynus new genus (see Systematic scope of this study, although we take limited Comments and appendix 7); Werneria Po­ actions to start this inevitable process. We che, 1903; ‘‘ Wolterstorffina ’’ Mertens, 1939 could place all of the names that are demon­ (see Systematic Comments). strably derived from ‘‘ Bufo ’’ into the syn­ CHARACTERIZATION AND DIAGNOSIS: Several onymy of Bufo , thereby providing a monoof the larval characters in our analysis (from phyletic taxonomy. However, because much Haas, 2003) optimize as synapomorphies of of this paraphyly was understood in 1972 Bufonidae : (1) diastema in larval lower lip (various papers in Blair, 1972a), it is clear papillation (Haas 9.1); (2) m. diaphragmato­ that social inertia is standing in the way of praecordialis absent (Haas 25.0); (3) lateral progress. We judge that progress will require fibers of m. subarcualis rectus II–IV invade the partition of ‘‘ Bufo ’’ into more informainterbranchial septum IV (Haas 29.1); (4) tive natural units.

processus anterolateralis of crista parotica A recent study on New World Bufo by absent (Haas 66.0); (5) larval lungs rudimen­ Pauly et al. (2004) had not appeared when tary or absent (Haas 133.0, also in Ascaphus we were designing our sampling strategy. and some Litoria ). Graybeal and Cannatella That work provides additional guidance in (1995) noted the fusion of the basal process our development of an improved taxonomy, of the palatoquadrate with the squamosal although the study differs from ours in ana­ (Baldauf, 1959), although they noted that not lytical methods and assumptions, taxon samenough taxa had been evaluated to ensure pling, and amount of data involved (2.5 kb

that this is the appropriate level of optimi­ of mtDNA in the study by Pauly et al., 2004, zation of this character. and ca. 3.7 kb/terminal of mtDNA and

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 215

in figure 50 and 60; the results of Pauly et Bufo asper group on the basis of morpholoal. (2004) are shown in figure 68 , and a com­ gy, while Liu et al. (2000) allied it with the parison of the taxa held in common by the B. melanostictus group on the basis of motwo studies is shown in figure 69 . Both stud­ lecular evidence. For the present we accept ies found the position of Bufo margaritifer its assignment to Duttaphrynus (the Bufo meto be remarkable. The difference is that we lanostictus group). Fei et al. (2005) regarded think further resolution should come from Torrentophryne to be part of this clade, a additional data and denser sampling rather conclusion not supported either by the study than from invoking one from among a re­ of Liu et al. (2000) or by our analysis, which stricted set of published models of molecular place Torrentophryne in Bufo (sensu stricto). evolution to explain the issue away. We restrict Phrynoidis to the Bufo asper

On the basis of our data analysis, as well group. We also suggest that some of the charas other information (e.g., Pauly et al., 2004), acters that optimize to this branch in our tree we can partition the following hypothesized (appendix 5) are synapomorphies of Phrymonophyletic units out of ‘‘ Bufo ’’ ( fig. 70 ): noidis.

(1) [476] Rhaebo Cope, 1862 (type spe­ (3) Rhinella Fitzinger, 1826: 39 (type species: Bufo haematiticus Cope, 1862 ). We rec­ cies by monotypy: Bufo [Oxyrhynchus] proognize the species of the Bufo guttatus boscideus Spix, 1824 ). We apply this name group, the sister group of all bufonids except to the Bufo margaritifer group (see appendix Melanophryniscus , Atelopus , Osorphophry­ 6 for nomenclatural comment and appendix ne, and Dendrophryniscus (see figs. 50, 60), 7 for content). The most recent morphologias Rhaebo (see appendix 5 for molecular cal characterization of the group (as the Bufo synapomorphies, appendix 6 for nomencla­ typhonius group) was by Duellman and tural comment, and appendix 7 for content) Schulte (1992), although their diagnoses exon the basis of their lack of cephalic crests, plicitly refer to overall similarity, not synatheir yellowish­orange skin secretions (white pomorphy. Hass et al.’s (1995) study of imin other nominal Bufo ; R.W. McDiarmid, munological distances found the group to be personal commun.), presence of an omoster­ monophyletic, but their outgroup samples num (otherwise found, among bufonids, only were limited to Bufo marinus and B. spinuin Nectophrynoides and Werneria [J.D. losus . Baldissera et al. (1999) provided evi­ Lynch , 1973: 146], Capensibufo [Grandison, dence (restricted to R. margaritifer ) from the 1981], and the Cranopsis valliceps group [J. nucleolar organizer region (NOR) that Rhi­ R. Mendelson, III, personal commun.]), and nella may be distantly related to Chaunus . hypertrophied testes (Blair, 1972c, 1972d), Most species of Rhinella have distinctive and which in combination differentiate Rhaebo extremely expanded postorbital crests in oldfrom all other bufonids. (See appendix 6 for er adult females, although this does not apnote under Bufonidae on this name.) pear to be the rule, so the diagnosis of the

(2) Phrynoidis Fitzinger, 1843: 32 (type group needs refinement. Should Rhinella Fitspecies : Bufo asper Gravenhorst, 1829 ). Be­ zinger, 1826, be found to be nested within cause it is more closely related to Pedostibes Chaunus Wagler, 1828 , the name Rhinella than to other ‘‘ Bufo ’’, we recognize the Bufo will take precedence for the inclusive group. asper group (see appendix 7 for content) as Given our taxon sampling, we cannot rule Phrynoidis . Inger (1972) provided morpho­ out the possibility that Rhinella and Rhamlogical differentia that serve to distinguish phophryne are not reciprocally monophylet­ Phrynoidis from other bufonid taxa. Which ic. However, although placing all the inof the suggested characters is synapomorphic volved species in a single genus would minis not obvious, and additional morphological imize the risk that we are wrong, we believe work is needed. Further, the monophyly of such caution to be counter­productive. Also, this taxon with respect to Pedostibes , and from a more pragmatic position, this would possibly to other unsampled genera, is an require all species currently placed in Rhamdi­

open question. The relationship of Bufo gal­ phophryne to be transferred to Rhinella eatus to this taxon is arguable. Dubois and based only on the possibility of nonmono­ Ohler (1999) provisionally allocated it to the phyly, not evidence. The genera may be 216 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297 2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 217

agnosed by the number and size of eggs fer from all species of Rhinella in possessing (many and small in Rhinella ; few and large a conspicuously elongate snout, reduced in Rhamphophryne ); by the adductor man­ number of vertebrae, and vocal sacs with slitdibulae musculature (both the m. adductor like openings. Likewise, most (but not all) mandibulae posterior subexternus and m. ad­ species of Rhinella differ from all species of ductor mandibulae externus superficialis [‘‘S Rhamphophryne in possessing protuberant 1 E’’ of Starrett in J.D. Lynch , 1986] present vertebral spines, greatly expanded alate postin Rhinella ; only m. adductor mandibulae orbital crests, and a leaf­like dorsal pattern.

subexternus [‘‘E’’ of Starrett in J.D. Lynch , Many questions remain regarding the re­ 1986] present in Rhamphophryne ); by the lationships between these and other New thigh musculature (m. adductor longus pre­ World bufonids. For example, Crepidophrysent in Rhinella , absent in Rhamphophryne ); ne epiotica possesses the same jaw muscuby liver morphology (trilobed with left side lature and liver morphology as Rhampholarger than right side in Rhinella , bilobed phryne and shares large, unpigmented eggs, with right side massive, conspicuously larger similar hand and foot morphology, and abthan left side in Rhamphophryne ); and by ex­ sence of the ear, suggesting it may be closely the

; tensively webbed hands and feet in Rham­ related to Rhamophophryne. However, phophryne (T. Grant, personal obs.). Most morphological results of Graybeal (1997 (but not all) species of Rhamphophryne dif­ data undisclosed) suggest that Crepidophry­

218 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297 2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 219

ne is imbedded within Cranopsis . Similarly, of the B. melanostictus group, although this, Andinophryne is characterized as possessing too, is not consistent with our molecular ev­ an omosternum, anteriorly ‘‘firmisternal’’ idence. and posteriorly ‘‘arciferal’’ pectoral girdle (5) Epidalea Cope, 1864 (type species: (for pectoral girdle morphology see Kaplan, Bufo calamita Laurenti, 1768 ) is available 2004), a complete ear, partially webbed for Bufo calamita (see appendix 6 for nohands, elongate paratoid glands, and lacking menclatural comment and appendix 7 for the m. adductor longus of the thigh (Hoog­ content). We had hoped to associate the name moed, 1989b). However, none of these char­ Epidalea Cope, 1864 , through its type ( Bufo acters is unique or clearly derived relative to calamita Laurenti, 1768 ), to the Bufo viridis likely relatives (e.g., Rhamophryne or Rhi­ group. However, association of Bufo calamnella ), and their relationships require further ita with the Bufo viridis group is seemingly investigation. based solely on overall similarity (Inger,

(4) [491] Ingerophrynus new genus (type 1972). The results based on DNA sequences species: Bufo biporcatus Gravenhorst, 1829 ; presented by Graybeal (1997; fig. 25) do not etymology: Robert F. Inger 1 Greek: phry­ place B. calamita and B. viridis as closest nos [toad]). This name commemorates the relatives. Because the phylogenetic evidence extensive contributions of Robert F. Inger to so far published (Graybeal, 1997) does not the herpetology of tropical Asia and the Sun­ suggest that B. calamita is a member of the das, as well as to the systematics of Asian B. viridis group (but see caveat regarding bufonids. The topology described by our ex­ Graybeal’s data in ‘‘Review of Current Taxemplars Bufo celebensis , B. galeatus , B. div­ onomy’’), we place them in separate genera ergens, and B. biporcatus suggests a major as an interim measure. We expect that, as buclade of tropical Asian bufonids. We pre­ fonid phylogenetics become better undersume that this clade contains all species of stood, the name Epidalea will be attached to the Bufo biporcatus group (see appendix 7 a larger group than just this one species. for content) in addition to B. celebensis Gün­ (6) Pseudepidalea new genus (type spether, 1859 ‘‘1858’’, and B. galeatus Günther , cies: Bufo viridis Laurenti, 1768 ; etymology: 1864. Inger (1972) provided differentia that in reference to the overall morphological distinguish the Bufo biporcatus group from similarity of members of the ‘‘ Bufo ’’ viridis the remaining Bufo , although it is not obvi­ group to Epidalea calamita ; see appendix 7 ous which of these characters are synapo­ for content). Liu et al. (2000) presented weak morphies. We also suggest that our branch evidence that Bufo raddei is not part of the 491 (see appendix 5) contains several molec­ Bufo viridis complex (their exemplars being ular synapomorphies that distinguish this Bufo oblongus danatensis and B. viridis ). For clade from all others. this reason we regard B. raddei as being only

Association of Bufo celebensis and B. gal­ provisionally assigned to this genus. A set of eatus with the Bufo biporcatus group as parts differentia provided by Martin (1972) will of Ingerophrynus rests entirely on molecular serve to distinguish this group from other buevidence (summarized in appendix 5), al­ fonid taxa, although, as in many such diagthough we expect that some of the characters nostic summaries, we cannot identify which that differentiate the B. biporcatus group characters are apomorphies and which are from other ‘‘ Bufo ’’ also apply to these two plesiomorphic. We do, however, suggest that species. Bufo celebensis had not previously the molecular synapomorphies provided in been associated with any other species of appendix 5 will serve to diagnose this taxon. Bufo , so our hypothesis of relationship is (7) Duttaphrynus new genus (type spenovel and suggests that Ingerophrynus sits cies: Bufo melanostictus Schneider, 1799 ; etastride Wallace’s Line. ymology: S.K. Dutta 1 Greek: phrynos

Dubois and Ohler (1999) provisionally al­ [toad]) reflects the contributions to herpetollocated B. galeatus to the B. asper group (5 ogy by Sushil Kumar Dutta, noted Indian for

by Phrynoidis ), but this allocation is not consis­ herpetologist). We erect this generic name tent with our molecular evidence. Liu et al. the Bufo melanostictus group as defined (2000) placed B. galeatus as the sister taxon Inger (1972) and subsequent authors. Al­

220 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

though decisive evidence for the monophyly letic Torrentophryne to be nested within the of Duttaphrynus is currently lacking, we hy­ otherwise monophyletic Bufo bufo group, pothesize that the group is monophyletic and which is consistent with our results. We suggest that detailed analysis of this group therefore follow Liu et al. (2000) in placing and close relatives will document this. Mor­ Torrentophryne in the synonymy of Bufo phological differentia provided by Inger (sensu stricto). Clearly, our taxon sampling (1972) serve to distinguish this group from is insufficient to allocate all species of reother ‘‘ Bufo ’’, although which characters are maining ‘‘ Bufo ’’ to identified clades and, as apomorphies and which are plesiomorphies we suggest later, we think that ‘‘ Bufo ’’ speremains unknown. We also suggest that at cies not allocated to this or other nominal least some of the molecular synapomorphies clades should be associated with this generic for ‘‘ Bufo ’’ melanostictus in our tree are syn­ name in quotation marks pending resolution apomorphies for Duttaphrynus (see appendix of their phylogeny. 5, for Bufo melanostictus ). (10) Vandijkophrynus new genus (type

(8) Peltophryne Fitzinger, 1843 (type spe­ species: Bufo angusticeps Smith, 1848 ; etycies: Bufo peltocephalus Tschudi, 1838 , by mology: E. Van Dijk 1 Greek: phrynos original designation) is a monophyletic ra­ [toad], commemorating Eddie Van Dijk, notdiation within nominal ‘‘ Bufo ’’ and was most ed South African herpetologist and indefatirecently synonymized with Bufo by Pramuk gable tadpole specialist). (See appendix 7 for (2000). In the most recent study of the rela­ content and new combinations.) We erect this tionships of this group, Pramuk (2000) ana­ genus for the Bufo angusticeps group as diflyzed morphological characters and mtDNA ferentiated by Tandy and Keith (1972; exsequence data and found Peltophryne (as the cluding Bufo / Capensibufo tradouwi and C. Bufo peltocephalus group) to be most closely rosei , which do not have the distinctive rerelated to the American Bufo granulosus ticulate dorsal pattern of the core group and group (see also Pregill, 1981; Pramuk, 2000; are placed phylogenetically far away in our Pramuk et al., 2001). Nevertheless, Pramuk analysis) and by Cunningham and Cherry (2000) rooted her cladogram on the Bufo re­ (2004). Our discovery of the exemplar of this gularis group and otherwise had relatively group, B. angusticeps , as the sister taxon of sparse outgroup taxon sampling. Our data in­ Stephopaedes is consistent with the results of dicate strongly that the Bufo peltocephalus Cunningham and Cherry (2004) . Should group is not closely related to the Bufo gran­ Vandijkophrynus be found to be synonymous ulosus group or any other American toad, but with Poyntonophrynus (see below), we select is, instead, the sister taxon of the African tax­ Vandijkophrynus to have priority under the on Schismaderma , which was not included in provisions of Article 24.2.1 (Rule of First previous studies of Peltophryne . The biogeo­ Revisor) of the International Code of Zoographic track suggested by this finding in­ logical Nomenclature (ICZN, 1999). vites further work. We therefore resurrect (11) Mertensophryne Tihen, 1960 (type Peltophryne (see appendix 7 for content) for species: Bufo micranotis rondoensis Loverthe Bufo peltocephalus group, as distantly re­ idge, 1942). We suggest that at least some of lated to other Neotropical toads. (See the no­ the molecular synapomorphies (appendix 5) menclatural comment in appendix 6.) that optimize to our Stephopaedes anotis are

(9) [499] Bufo Laurenti, 1768 (type spe­ synapomorphies for a larger Mertensophrycies: Rana vulgaris Laurenti, 1768 , by sub­ ne. Complicating discussion of phylogeny in sequent designation of Tschudi, 1838: 50). the vicinity of Stephopaedes is Mertenso­ We restrict the generic name Bufo (sensu phryne and the Bufo taitanus group (see apstricto) to the monophyletic Bufo bufo group pendix 7 for content), which is a group of of Inger (1972) and subsequent authors (see African toads that lack tympani and coluappendix 7 for content). Inger (1972) sug­ mellae; that frequently show digit reduction gested morphological differentia for this tax­ (Tandy and Keith, 1972); and that have been

on that separate it from other bufonid taxa, been suggested to be related to Capensibufo although their polarity remains to be docu­ (Tandy and Keith, 1972). Graybeal and Canmented. Liu et al. (2000) found a paraphy­ natella (1995) and Graybeal (1997) suggest­

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE 221

ed that Stephopaedes and Mertensophryne (2004). Poytonophrynus is characterized by are nested within at least some component of lacking a tarsal fold (a presumed apomorthe B. taitanus group; Cunningham and phy), having parotoid glands indistinct and Cherry (2004) arrived at similar conclusions. flattened (Poynton, 1964a), and the tympa­ Müller et al. (2005) described the tadpole of num being small but distinct (Tandy and B. taitanus and reported that it has a crown Keith, 1972). We did not study any member that encircles the eyes as in Stephopaedes but of this group, but on the basis of the DNA is not so well developed. In no studies have sequence results presented by Cunningham the relationships of Mertensophryne , Stepho­ and Cherry (2004), it is a monophyletic paedes, and the Bufo taitanus group been group, closely related to Mertensophryne evaluated with adequate taxon sampling, and (sensu lato) and Vandijkophrynus . See apthe questions of relationship have remained pendix 7 for content and new combinations. recognized (e.g., Tandy and Keith, 1972) but See appendix 7 for content and new combiunresolved for more than 30 years. What is nations.

known is that the Bufo taitanus group, Mer­ (13) [506] Amietophrynus new genus tensophryne, and Stephopaedes lack colu­ (type species: Bufo regularis Reuss, 1833 ; etmellae (convergently in the clades composed ymology: Jean­Louis Amiet, an influential of [1] Wolterstorffina , Werneria , Nectophry­ herpetologist of West Africa, 1 Greek: phryne, and likely Laurentophryne [Tihen, 1960; nos [toad]). We erect this taxon for all Afri­ Grandison, 1981]; [2] Didynamipus ; and [3] can 20­chromosome ‘‘ Bufo ’’ discussed by Capensibufo rosei ), and likely form a mono­ Cunningham and Cherry (2004; the clade phyletic group. Furthermore, Stephopaedes , subtended by our branch 506), as well as the Mertensophryne , and at least one member of 22­chromosome ‘‘ Bufo ’’ imbedded within the Bufo taitanus group ( B. taitanus ; H. this clade (the Bufo pardalis group of Cun­ Müller et al., 2005) have accessory respira­ ningham and Cherry, 2004). This includes tory structures on the head of the larva. toads previously included by various authors (Nevertheless, differences among these struc­ in the Bufo blanfordi group, B. funereus tures suggest the possibility of nonhomolo­ group, B. kerinyagae group, B. latifrons gy; Dubois, 1987 ‘‘1985’’; Poynton and group, B. lemairii group, B. maculatus Broadley, 1988 .) Mertensophryne is current­ group, B. pardalis group, B. perreti group, ly monotypic, and Stephopaedes contains B. regularis group, B. superciliaris group, three species. Our action to promote further and B. tuberosus group. Although at least research is to place the Bufo taitanus group, some of these groups are monophyletic, we Mertensophryne , and Stephopaedes into an do not recognize species groups within Amieenlarged Mertensophryne , retaining Stepho­ tophrynus at this time, because several of the paedes as a subgenus, in order not to lose existing groups are monotypic (e.g., B. lerecognition of this monophyletic group. (See mairii) or clearly nonmonophyletic (e.g., B. appendix 7 for new combinations.) Loss of regularis group). We think that recognition the middle ear is synapomorphic at this level of species groups should follow a more and, although larvae are unknown for several densely sampled study of the Amietophrynus members of the Bufo taitanus group, we sus­ and near relatives. Although not previously pect that the accessory repiratory structures suggested to be a member of the 20­chroon the head of larvae is a synapomorphy as mosome group, our phylogenetic results alwell. Ongoing work by Channing and col­ low us to predict that Bufo tuberosus is a 20­ laborators will address this further. chromosome frog.

(12) Poyntonophrynus new genus (type Beyond the 20­chromosome condition that species: Bufo vertebralis Smith, 1848 ; ety­ is apomorphic for group (Bogart, 1968; Cunmology: J.C. Poyton [commemorating John ningham and Cherry, 2004), molecular trans­ C. Poynton, noted South African herpetolo­ formations diagnose the taxon unambiguousgist] 1 phrynos [Greek: toad]). We recognize ly (see appendix 5). Moreover, the monophy­

Figures

Fig. 9. Tree of Plethodontidae by Mueller et al. (2004), with the traditional taxonomic assignments (Desmognathinae 1 tribes of Plethodontinae; Wake, 1966) placed on the right, with taxonomic fragments numbered for clarity. The generic taxonomy was updated to reflect name changes of former Salamandra luschani to Lyciasalamandera (Veith and Steinfartz, 2004) and Hydromantes italicus to Speleomantes. The results reflect a Bayesian analysis of entire mt DNA genomes (number of informative sites not stated, but analyzed fragments totalled 14,040 bp), with control region and ambiguously alignable region excluded. Sequences were aligned with default costs of GCG v. 10.3 (Accelrys, San Diego; cost of 8 for gap creation and extension cost of 2) and subsequently adjusted manually. It was not stated whether gaps were treated as evidence or as missing data.

Fig. 68. Maximum­likelihood tree of predominantly New World Bufonidae suggested by Pauly et al. (2004) on the basis of 2,370 bp (730 informative sites) of mitochondrial DNA (12S, tRNAVal, and 16S). Alignment was done under Clustal (Thompson et al., 1997; cost functions not disclosed) then modified manually. Gaps were considered to be missing data and the substitution model assumed for the maximum­likelihood analysis was GTR 1 G1 I.

Fig. 69. Comparison of our bufonid parsimony results, via terminals held in common (see fig. 50, 60) with those of Pauly et al. (2004) (fig. 68). Taxa whose relative placement differs substantially between the two studies are in boldface.

Fig. 70. Generic changes suggested for bufonid taxa that we studied. This figure shows our terminals